the impact and effectiveness of emergency nutrition and ... · therapeutic nutrition programmes and...

The impact and effectiveness of emergencynutrition and nutrition-related interventions:a review of published evidence

1

Emergency, Nutrition Network Canadian International

Development Agency

Agence canadienne de

développement internationala

2004-2010

This review was undertaken by Andrew Hall and Becky Blankson (BB), Centre for PublicHealth Nutrition, School of Life Sciences, University of Westminster, 115 New CavendishStreet, London W1W 6UW, U.K and Jeremy Shoham (JS), Emergency Nutrition Network, 32Leopold Street, Oxford, OX4 1TW, U.K.

Production editing by Marie McGrath, ENN and report layout and design by Orna O'Reilly,www.holytrousers.com

This review was funded by the Canadian International Development Agency. June 2011.

Recommended citation: The impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidence 2004-2010. Hall A, Blankson B andShoham J. Emergency Nutrition Network, Oxford, UK. June 2011.

Front cover photos: D Doledec, Uganda, 2008; Veronique Priem, Niger, 2002; Mesfin Teklu, WV, Ethiopia, 2006; Mulugeta Wtsadik/UNHCR, Ethiopia, 2009;M Manary, Malawi, 2003.Back cover photos:© IRD - Sokhna, Cheikh; Hailu Sitotaw, Ethiopia, 2010; Valid Nutrition, 2009; WFP/S37, Kenya, 2008.

The impact and effectiveness of emergency nutritionand nutrition-related interventions: a review of published evidence (2004-2010)

3

Contents

1 Emergencies and interventions 4 1.1 Causes of emergencies 4

1.2 Main interventions during emergencies 4

1.3 Consequences of disasters 5

1.4 Determining the causes of malnutrition 5

1.5 Measuring the impact of interventions 5

2 Providing plausible evidence of impact 7 2.1 Comparison with historical data 7

2.2 Cohort studies 7

2.3 Natural experiments 1: dose-response effects 8

2.4 Natural experiments 2: case-control studies 8

2.5 Actual experiments: comparing interventions 8

2.6 Actual experiments, but not done during an emergency 8

3 Efficacy and effectiveness 9

4 Outcome measures 10

5 Methods 12

6 Results 13 6.1 General studies and surveys 13

6.2 The treatment of moderate and severe acute malnutrition 14

6.3 Micronutrients distribution 18

6.4 Food security or agricultural interventions 20

6.5 General ration distribution programmes e.g. food-for-work 20

6.6 Actual experiements: non-emergency 21

6.7 Costs and cost-efficiency 21

7 Discussion 22

8 Recommendations 25

9 Annexes 27

10 References 64

4

In the context of humanitarian emergencies, the term in-tervention describes any thing provided, or measure putin place, to prevent, mitigate, alleviate, treat or repair theharmful effects of whatever has caused the emergency.

1.1 Causes of emergencies

An emergency could be caused by:· A sudden natural physical event such as a flood, earthquake, tsunami, cyclone or hurricane.· A chronic natural event such as a drought, out-break of pests or disease that kills or injures crops or livestock upon which people depend for food or income.· An epidemic of disease that debilitates or kills large numbers of people.· A breakdown of order, institutions and supplies due to public unrest, civil war, ethnic conflict, government collapse or international wars· A political or economic crisis.

When an acute event occurs, it is relatively easy to definean emergency, although the scale and number of people in-volved will likely influence whether an official emergency isdeclared by a government or the United Nations. Howeverthe threshold between a slowly deteriorating situation andan emergency is hard to define in some circumstances, suchas a drought.

As the focus of this review is on interventions that couldimprove nutrition, the prevalence of wasting or underweightcould offer thresholds to define an emergency or disaster. TheWorld Health Organisation (WHO) consider that a prevalenceof wasting of ≥15% is ‘critical’1 but this can be exceeded rela-tively easily in countries such as India, Bangladesh andEthiopia where there is chronic and often seasonal undernu-trition. National governments, such as Ethiopia, may use theprevalence of wasting plus other criteria, sometimes de-scribed as ‘aggravating factors’, which could make the situa-tion worse to classify the severity of the situation.2

So the threshold between an emergency situation and achronically malnourished population is unclear and meansthat deciding whether an intervention is being applied in anemergency or not, is also often unclear.

1.2 Main interventions during emergencies

The main interventions provided by governments and reliefagencies during an emergency are:· Money, such as cash (conditional or not), credit or loans, or by buying livestock, called destocking and restocking.· Food, such as a general ration, blanket feeding for vulnerable groups, and supplementary and therapeutic food for young children.· Preventive health care, such as vaccinations, vitamin A

Emergencies andinterventions1

The impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidenceEmergencies and interventions

M Manary, Malawi, 2003.

Emergencies and interventionsThe impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidence

5

capsules and other micronutrients, iodised table salt, deworming, and impregnated bed nets.· Support for breastfeeding or artificial feeding if necessary.· Psychosocial care and counselling, especially in disasters that may have killed many people.· Water or water purification facilities or chemical sterilising tablets.· Sanitation facilities.· Shelter, clothes and domestic equipment.· Agricultural supplies, seeds, saplings and livestock· Rescue, relocation and physical safety, usually in the short term.

1.3 Consequences of disasters

A major focus of efforts during and after an emegency are:· To prevent deaths.· To sustain optimal feeding practices and prevent or minimise practices that put children at risk.· To treat and prevent undernutrition, particularly among young children.· To restore housing, infrastructure, health and education facilities, often by ‘building back better’· To restore or create livelihoods.· To increase robustness to future shocks.

1.4 Determining the causes of malnutrition

There is a common but mistaken assumption that cross-sec-tional data from surveys can be used to determine the causesof malnutrition. Data collected during surveys may be analysedto identify statistical associations between potential risk factorsand indices of anthropometric or nutritional status, but an as-sociation does not prove causation, for two main reasons.

First, the relationship could be confounded, in which boththe risk factor and the outcome variable are not relatedthrough a causal pathway but are both independently relatedto a third, confounding factor that is causative. One particularform is called confounding by intention. For example, amother may stop breastfeeding a child who is failing to gainweight, perhaps because of illness, so the wasting is mistak-enly associated with a lack of breastmilk rather than with theactual cause.

Second, cross-sectional data can also be mistakenly inter-preted because it cannot be determined which came first, theexposure or the outcome, called reverse causality bias. Forexample, if malnourished children are found to be infectedwith worms it cannot be known if the worms contributed tothe malnutrition or whether malnutrition predisposed to thechildren to worms. One of the most important principles fordetermining causality, that the exposure must precede theoutcome,3 is not met in cross-sectional data.

The situation is often made worse during a disaster be-cause new or more serious causes of malnutrition occur, suchas epidemics of disease or severe food shortages, which addto and exacerbate pre-existing malnutrition. The best solution

is to undertake a needs assessment based on an understand-ing of the potential causes of malnutrition, perhaps using atool such as the UNICEF conceptual framework to create achecklist. For these reasons, data collected during a disastercan rarely be used to determine the chronic, underlyingcauses of malnutrition in any community, but can be used toassess people’s immediate needs based on what they lack.

1.5 Measuring the impact of interventions

The focus of the first review4 and of this update is on findingand summarising evidence for the impact and cost-effective-ness of interventions provided during emergencies. There areseveral reasons why obtaining such evidence is conceptuallyand practically difficult.

First, because it is hard to show that something has beenprevented or has not happened. If deaths have been pre-vented for example, it is not possible to know, for obvious rea-sons, how many would have occurred without theintervention. The best that can be done is to show that thedeath rate has not increased. This requires prior data on thenormal death rate in the population before the emergency,which may not be available or, if it is known, may have beenunacceptably high in the first place. Any other estimate isspeculation, perhaps based on the death rates observed dur-ing other similar emergencies, although making such com-parisons can require many assumptions.

Second, to be able to assess the attributable benefit of anyintervention on human health such as nutritional status, it isnecessary to know what would have happened without it.The attributable benefit is the effect of an intervention on anoutcome that is additional to the change that would have oc-curred without the intervention. It is the antonym of attrib-utable risk.5 The usual conclusion in circumstances in whichhumanitarian aid is provided is to assume that any change inan outcome in the right direction is due to the interventionor, in epidemiological terms, the exposure. This is illustratedin Figure 1A. Figures 1B, 1C and 1D illustrate other possibleway to mis-estimate the attributable benefit, of which 1D isthe most worrying and important; if there is no change in ameasured outcome it could be assumed that the interventionhas had no effect when, in fact, it has been successful at pre-venting the situation from getting worse. The changes thatoccur at the same time as the intervention could be due toconcurrent interventions and activities, which is often the caseduring an emergency, or could be due to changes in externalfactors, such as climate and rainfall, which influence normalmethods of food production.

It is also possible that an intervention may do harm. Astudy after an earthquake in Indonesia reported that childrenwho were given donated breast milk substitutes had a greaterrisk of diarrhoea that children who did not.6

Figure 1 shows that estimating a simple change in meas-ures at two points in time is not sufficient to conclude that allor part of the impact is attributable to the intervention. Inorder to do this, a comparison group is needed to assess the

6

change that would have happened anyway, without the in-tervention, and thus estimate a difference between twogroups, not simply a change in one group. This is becauseother factors often influence nutritional status concurrently.For example if rainfall after a drought coincides with the dis-tribution of rations of food, the rain may lead to more fodderfor livestock which provide milk to give children a better diet,in addition to the ration, so Figure 1C may apply. For this rea-son the control or comparison group needs to be studied con-currently and in the same physical location, so is exposed tothe same external influences. To be able to measure the extra or attributable benefit andachieve internally valid results, the methods of evaluationshould ideally include:· Random allocation of subjects to groups to receive the intervention or not, so that any differences or confounding factors are evenly distributed between groups and are controlled for; prior stratification can be used to ensure that likely confounders are evenly distributed between groups.· Avoidance or minimisation of contamination between groups, so that only the intervention group gets the intervention.

· Clear, measurable and relevant outcomes that are known to be influenced by the intervention so that the difference in benefit between the groups can be estimated.· A sufficient period to detect differences between groups, because different outcomes may take different amounts of time to diverge.· Measures to ensure data quality including accuracy and precision.· An estimate of the probability that the difference is due to the intervention, not chance.

These criteria are almost impossible to meet during a hu-manitarian emergency for ethical reasons, mainly because itis not possible deliberately to withhold an intervention inorder to achieve a comparison or control group. This meansthat the evidence of a beneficial impact can rarely be ob-tained experimentally, usually it can only be derived from datathat provide plausible evidence of an effect.7 Plausible is de-fined as apparently true or reasonable, although what is rea-sonable to one person may not be to another. A cynic coulddefine plausible as an outcome that meets expectations, andif it meets the expectations of more than one person it couldbe concluded as proof. 8 Plausibility is the antithesis of prob-ability, as it is not numerical, it is subjective.

A. Mistaken conclusion 1: The attributable benefit isestimated by measuring the difference in prevalencebetween T0 and T1 which is assumed to be due to theintervention because it has fallen.

B. Mistaken conclusion 2: The attributable benefit isunderestimated because the prevalence of wastingwould have increased without the intervention.

D. Mistaken conclusion 4: The attributable benefit wasnot estimated at all because the intervention preventedthe prevalence of wasting from increasing, so no changewas observed.

C. Mistaken conclusion 3: The attributable benefit isover-estimated because the nutritional situation improved concurrently and the prevalence of wastingwould have fallen anyway.

Prev

alen

ce o

f was

ting

40

35

30

25

20

15

10

5

0

Assumption

Time 0 Time 1

InterventionΔ = 10%

Prev

alen

ce o

f was

ting

40

35

30

25

20

15

10

5

0

Comparison

Time 0 Time 1

Intervention

Δ = 15%

Figure 1. Examples of how the attributable benefit of an intervention can be mistakenly estimated using an example ofthe prevalence of wasting, a common target of nutritional interventions during an emergency, measured twice, over aperiod of time. Also shown is what might have happened to concurrent comparison group. The value of delta Δ is theattributable benefit under each different circumstance.

Prev

alen

ce o

f was

ting

40

35

30

25

20

15

10

5

0

Comparison

Time 0 Time 1

InterventionΔ = 5%

Prev

alen

ce o

f was

ting

40

35

30

25

20

15

10

5

0

Comparison

Time 0 Time 1

InterventionΔ = 10%

The impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidence Emergencies and interventions

The impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidence

7

Providing plausibleevidence of impact

There are a number of possible ways plausible evi-dence of an effect may be obtained. There may bemore than are proposed here; this is a topic currentlybeing developed by the Emergency Nutrition Net-

work (ENN)I.

2.1 Comparison with historical data

If data on key outcomes such as death rates, the prevalence ofanthropometric indicators, or micronutrient status have beencollected periodically - in good times as well as bad - then anincrease followed by a prompt return to normal after an inter-vention could plausibly be due to the intervention. Routinesurveillance data are currently collected in countries such asEthiopia, where there have been many emergencies, to pro-vide an early warning of an impending crisis, so they could suf-fice if key outcome indicators (see below) are measured. Anexample of this is a paper identified during the present review,of data from 900 surveys in the Eritrea, Ethiopia, Kenya, Soma-lia, Sudan and Uganda which recorded fluctuations in theprevalence of wasting.9 Such periodic data can indicate acause and effect relationship between an intervention and im-provement of the situation, provided that they are specific tothe circumstances in which the disaster occurs and to the eth-nic group and type of livelihood. Comparing data betweendifferent communities or with national averages may not bevalid because, paradoxically, the average is not common; mostcommunities are either above or below the average.

A major problem arises because of seasonality. Becauseagricultural production and malnutrition are both seasonaland tend to be cyclical, and because interventions may beprovided when the situation is at the worst point in a naturalcycle, the improvement in an outcome measure may have oc-curred anyway. Such data alone are weak and prior data maynot exist in places that rarely experience an emergency orcannot afford to collect them.

Another problem arises with interpreting longitudinal datawhen the intervention is widespread or comprehensive, suchas a change in policy that affects all individuals within a pop-ulation. There are, however, statistical approaches to estimat-ing impact that can control for bias.10

2.2 Cohort studies

If subjects enrol for interventions over a period of time whilean emergency continues, it is possible to use the later recruitsto provide evidence of what might have happened withoutany intervention. Early recruits may also benefit to a greaterdegree than later recruits if they are studied as a cohort (see3). However later recruits may be drawn from previously dis-advantaged communities within the population, so mighthave been different prior to the emergency. It cannot neces-sarily be assumed that all potential recruits are homogenouslymixed and then sampled over time, which is necessary toachieve equality in risk.

2

The impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidenceProviding plausible evidence of impact

Andy Catley, Ethiopia, 2005

8

2.3 Natural experiments 1 Dose-response effects

In most circumstances, there is not a uniform delivery or cov-erage of an intervention, or there are circumstances withinhouseholds that affect the implementation of the interven-tion. These differences or ‘variance’ create the circumstancesfor natural experiments in which measures of the effective-ness of delivery can be related to the degree of change in anoutcome. If the variance in the exposure is associated withthe variance in the outcome, there may be a causal associa-tion, although confounding can also occur. Again this re-quires paired data from a cohort of subjects so that themagnitude of a change in an outcome can be measured, suchas weight, height, mid upper arm circumference (MUAC) orhaemoglobin concentration, for example.

After the flood in Pakistan in 2010, each household wasable to claim a standard amount of food every month basedon a family of five people, two adults and three children. Itcould be hypothesised that small households would be atlower risk of having malnourished children than large house-holds after controlling for prior socio-demographic indicators.

There is also the potential to examine how changes in nu-tritional outcomes are related to initial characteristics. For ex-ample, weight gain is likely to be related to initial body weightand to the degree of undernutrition of children. This can bestandardised by expressing weight gain in terms of grams perkilogram of initial body weight per day, or changes in heightor MUAC in mm/day (see below).

2.4 Natural experiments 2 Case-control studies

If data on changes due to interventions are needed, then it isbest to study cohorts of individuals so that paired data are ob-tained so, for example, weight gain can be estimated. To ob-tain paired data requires that that children are identified onat least two occasions, something that can be difficult to doreliably in emergencies.11 The alternative is to study large in-dependent, random samples at intervals, but the potential forapplying analytical statistics is weaker than if paired data areobtained.

Paired data allow nested case-control studies, eitherprospective or more likely, retrospective, in which parametersthat capture the efficiency of an intervention can be relatedto a standardised change in an outcome measure. For exam-ple, during a period of severe food shortage in northernKenya, a blanket feeding programme was implemented foryoung children in which the mother was given enough corn-soy blend and oil each month to provide a child with an aver-age of 1,000 kcal/d. It was hypothesised that theanthropometric status of a singleton child would be betterthan a child with older siblings, with whom the ration mightbe shared. In such an analysis it would be best to control forage, sex and the date of recruitment, but a sample of severalhundred children in many villages could have helped to con-trol for environmental and secular changes that occurred at

the same time. It would have to be assumed that the parentsdid not eat the ration, too.

2.5 Actual experiments: Comparing interventions

When it is not known whether one intervention is better thananother, but both are known to be effective, it can be ethicalto undertake a comparative study, especially if the study sub-jects are relatively low risk groups. For example, moderatelyundernourished children could be given corn soy blend (CSB)and oil or ready-to-use food (RUF).. If the subjects are ran-domly allocated, if informed consent is obtained and, ideally,if the person making the anthropometric measurements doesnot know which child is receiving which ration to introducean element of blinding, then data on the comparative effec-tiveness of the treatments can be obtained. Such allocationscan be done by individuals or in clusters, such as villages ortreatment centres, providing that there are a sufficient num-ber of clusters. Establishing the number of clusters requiresprior data on the intra-cluster correlation coefficient for theoutcome variable, which may not be available, so an arbitraryrule of thumb of a minimum of 25 clusters may be applied.

Such experiments can rarely be planned in detail, so re-quire preparedness in principle: to have potential interven-tions in mind, to assess whether the circumstances will allowrandom allocation, and to be able to get ethical approval rap-idly from a recognised Review Board. Without ethical ap-proval it will not be possible to publish the resulting data.

2.6 Actual experiements: non-emergency

For many practical reasons, experimental studies need carefulplanning and ethical approval, which takes time, so cannotbe set up within the short notice in which an emergency ornatural disaster occurs. For this reason, much of the dataabout the effectiveness of the interventions that may be ap-plied during an emergency actually come from managed, in-dependent, scientific studies. The data may be internally validfor the community in which the study was done, but are theyvalid in other circumstances, such as during an emergency?Again this is where plausibility must be considered. Is it rea-sonable to assume that a vitamin A capsule given during anemergency is as efficacious as one given as a part of nationalvitamin A or health day? There are two main reasons whythere may be differences: endogenous reasons due to geno-typic and phenotypic differences between individuals, whichare beyond the scope of this review, and exogenous reasonsdue to differences in the delivery of interventions that affectcoverage of the population at risk.

The impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidence Providing plausible evidence of impact


9

The main exogenous difference between individualsduring an emergency lies in the distinction betweenthe efficacy and effectiveness of an intervention. Ef-ficacy is the attributable benefit of an intervention

given under ideal circumstances while effectiveness is the at-tributable benefit of an intervention under usual circum-stances. The difference between the two is usually estimatedas coverage or the efficiency of allocation or whatever meas-ures are put in place to make sure that beneficiaries actuallybenefit to the maximum possible degree from the interven-tion. So for vitamin A, a crucial difference between an emer-gency and normal circumstances is the coverage oftreatment, which may be greater than normal during anemergency because beneficiaries are more easily accessiblein a camp, for example. So if there is variation in coverage,then there may be a proportional variation in outcomes as aconsequence, so a natural experiment may have taken place.The main question is whether the causal pathway was director indirect. But, whatever the causal pathway, the measure-ment of outcomes is crucial to the estimation of any impact,whether attributable and probable or simply just plausible.

Efficacy andeffectiveness3

The impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidenceEfficacy and effectiveness

© IRD - Sokhna, Cheikh

The impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidence Outcome measures

10

Outcomemeasures4

The main expected effects of interventions duringemergencies can be grouped as follows:

· To increase money and assets, estimated as in come or monetary value, or by the presence or absence of capital items.· To increase food security, estimated as agricultural production, dietary diversity and food intake· To increase anthropometric status, which is estimated from measurements of weight, height and age which are compared with values for a well-nourished reference population and expressed in standard deviations (z-scores) or as the percentage of children with values less than -2 or -3 z-scores.· To increase nutritional status, measured as the concentration in blood and body fluids of micronutrients or indicative substances such as haemoglobin.· To sustain breastfeeding and ensure the dietary diversity of complementary foods.· To improve health, estimated as the incidence of infectious diseases.· To prevent deaths.· To improve mental and psychological status and reduce stress, measured by psychometric tests or by the concentration of hormones in blood.

A very important outcome of interest is weight gain, as thiscan be achieved quickly by a number of different interven-tions, although most quickly by giving nutrients, and becauseimproving nutritional status greatly reduces the risk of dying.The rate of weight gain is related to the following:· Age and initial body weight, as the size of children influences how much they can eat and thus gain each day.· The initial degree of undernutrition, as the most undernourished children have more weight to catch up.· The sex of the child, although this effect is small in very young children.· The ethnic origin of the child, although they are rarely mixed or diverse to a degree that leads to differences in measured outcomes.· The amount and density of energy and nutrients consumed.· The period of treatment, as weight gain tends to be greatest initially during recovery and then slows down over time.12

For these reasons weight gain is usually expressed asgrams per kilogram of initial body weight per day (g/kg/d org.kg-1.d-1) and is often estimated over the first 28 days (4weeks) of treatment when the weight gain is likely to be mostrapid, and for non-oedematous children only. This is a some-

Stuart Katwikirize/C-SAFE, Lesotho, 2006

Outcome measuresThe impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidence

11

what arbitrary period but is long enough to achieve a satis-factory weight gain and represents a typical period of treat-ment and a sufficient period of recovery before discharge. Thetherapeutic or supplementary foods given to young childrentypically provide between 75 and 210 kcal/kg/d.

Weight gain may also be affected by the presence ofoedema, which will tend to mask low body weight. The pres-ence of oedema means that z-scores of weight-for-height andweight-for-age are underestimated, or that weight gain canonly accurately be estimated once the oedema has gone, orthat weight gain has to be calculated separately for childrenwith or without oedema.

The criteria used to admit children to a feeding pro-gramme can also have an effect on measuring the outcomeof an intervention. This may happen especially if resourcesare in limited supply so that priority is given to the most un-dernourished children.

If different criteria are used to admit children, such asMUAC <110 mm or <80% of the median or weight-for-height,then such children are not necessarily similarly undernour-ished, and if both criteria are applied it is not clear how to clas-sify a child who exceeds one criterion but not another. 13

Assessing the criteria of success or discharge from a pro-gramme, an alternative estimate of the success of a pro-gramme, is also variable and inconsistently defined. This canbe done either by estimating the time it takes for children tomeet the stated criteria or by assessing the percentage of sub-jects who meet the criteria within a given period. The mainproblems lie in defining and standardising the criteria for dis-charge and the fact that not all children are equally under-nourished, so some may meet the discharge criteria morequickly than others. This makes comparisons difficult be-tween different samples of children, especially if several dif-ferent criteria need to be met. Some examples are:· Loss of oedema · Percentage weight gain · Crossing a threshold MUAC e.g. 115 mm· A combination of MUAC and an increase in weight 13

· Achieving a weight-for-height of >-2 z-scores14

· Achieving for at least 2 weeks a median weight-for- height of 85% of the median15

· Reaching 100% of weight-for-height.16

As it can be seen, there are no standard criteria but averagedaily weight gain over a period of 28 days and standardisedfor initial body weight offers the best method to compare theefficacy or effectiveness of different treatments. The first finaledition of the SPHERE humanitarian standards published in2002 (reported by Collins et al. in 200217) apparently specifiedthat the minimum mean rate of weight gain should be8g/kg/d, but no rate is specified in the 2011 edition.18

Similar data can be used to assess the effect of food on lin-ear growth but, as gaining length or height usually takeslonger than gaining weight, the rate is usually measured in

mm per day over a period of 8 weeks and is not standardisedfor initial height.

This review will now focus on the recently published evi-dence of the effectiveness of interventions in emergenciessince the review of Duffield et al. (2004).

12

Methods5

Asearch of the published literature was undertakenusing the PubMed and Cochrane Databases withlimits set for papers published since 2004, the yearof the first review. The searches were done be-

tween November 2010 and January 2011. The search termsused were the same as those given in Appendix 3 of Duffieldet al. (2004). Papers were classified into seven categories:

1. General studies and surveys.2. Treatment of moderate or severe acute malnutrition separately, as community-based treatment or as hospital or clinic-based treatment.3. Micronutrients distribution.4. Food security or agricultural interventions5. General ration distribution programmes e.g. food-for- work, food-for-cash.6. Conditional cash transfers, loans or vouchers.7. Cost and cost-efficiency.

The papers were then reviewed by JS and BB and thoseconsidered not to be relevant were dropped. The papers wereclassified into those relating to emergencies and those of rel-evance but not done during emergencies. A summary of thepapers is provided as a separate documentII. This review sum-marises the papers concerning emergencies only.

The analysis of data was undertaken using Stata version 11and graphs were produced using Microsoft Excel.

The impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidenceMethods

Hailu Sitotaw, Ethiopia, 2010


13

Results6

6.1 General studies and surveys

The prevalence of undernutrition of young Sri Lankan chil-dren living in relief camps a month after the Asian tsunami of2004 was reported to be significantly higher than the SriLankan national average,19 but it might have been the samebefore the disaster. This illustrates the value of having priordata.

Surveys have also been reported in Mozambique20 (wherethere was also no prior data), Afghanistan21 (to describe and‘understand’ the causes of malnutrition in an area after con-flict); and in four camps for internally displaced people inUganda22 (to describe the situation and factors associatedwith malnutrition).

Surveys of the prevalence of wasting over a period of 3-4years at seven sites in the Congo after a civil war indicatedthat the prevalence had declined by 1.3 to 10.4%.23 Thesenumbers were extracted from Figure 2 in the paper, so are es-timates. Data at one site (Shabunda 02) is uncharacteristic.No confidence intervals were given so no assessment can bemade about the statistical significance of any differences.

A natural experiment was reported in a retrospective studyof changes in the composition of general food rations inBhutanese refugee camps in Nepal and subsequent changes

in mean birth weight and the incidence of low birth weight.24

It was suggested that the fall in the incidence of low birthweight from 16% to 8% between 1996 and 1998 and the in-crease in mean birth weight was due to a change in the foodration, which provided more micronutrients. The rates of lowbirth weight in the camp in 1998 were lower than in bothBhutan and in Nepal. The fact that all refugees received thesame ration and the lack of a concurrent control group meansthat the cause of the apparent change cannot be determined. An evaluation of the impact of nutrition care centres estab-lished in Baran District, Rajasthan, India during a severedrought reported a reduction in prevalence of underweightfrom 66.7% to 59.6% after six months and a reduction in theprevalence of severe underweight from 32.9% to 26.1%.25 Noconfidence intervals were given around estimates but theycan be estimated from the data to be around ± 2.3% for bothestimates so were: 66.7 % (64.4, 69.0) down to 59.6% (57.3,61.9), so the smallest possible difference between the twopercentages is about 2.5%, which is a small change in preva-lence of underweight. There was no control group. The effectof the programme, which is described in great detail and wasapparently widely replicated, was considered to be ‘positive’.The authors stated that the standardisation of procedures had‘been worked out ‘to deal with the problem of child malnutri-tion’ (p 205) 25 but actually had little effect.

The impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidenceResults

Ellen van der helden/MSF Ethiopia

The impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidence Results

14

Conclusions

Data from surveys are useful to describe the situation for thepurposes of advocacy and if data are collected after the emer-gency has ended, they can show that the situation is now bet-ter, although this assumes that the sample of individualsstudied on both occasions has similar underlying character-istics. One problem that may occur with survey data is that inan emergency the people who are better-off tend may leavethe area so that the residual population tends to be the poor-est and most deprived. This means that the prevalence of un-dernutrition may increase because the denominator haschanged, as well as increasing because of the emergency.

The 2011 SPHERE standards only specify that z-scores ofweight-for-height should be calculated using the WHO refer-ence values, measurements of MUAC should be taken, andthey should be used to classify children based on thresholdsof one or both measurements.18 There are advantages to lim-iting data to body measurements only and avoiding the prob-lems with determining a child’s age. A recent evaluation of ablanket feeding programme in Kenya indicated that motherswere under-estimating their child’s date of birth, perhaps toensure that they were eligible to get a ration even though thebasis of inclusion was height, not age.11 But data on age allowweight-for-age, height-for-age and Body Mass Index (BMI)-for-age to be calculated and if the composite index of anthro-pometric failure is calculated, then the total burden ofundernutrition can be estimated.26

6.2 The treatment of moderate and severe malnutrition

These are dealt with together because they are not separateconditions even though they lead to separate programmes:children who are classified as severely wasted are a subset ofall children who are classified as wasted as both are based onthe same index, weight-for-height. The most important development in the field since the lastreview is the increasing use of RUFs, in addition to fortifiedblended food, which has greatly improved the treatment ofwasting and is now being used in supplementary as well astherapeutic nutrition programmes and in community as wellas clinic-based programmes.

Ready to use therapeutic food (RUTF) is a type of RUF thatis an energy-dense soft paste made of ground peanuts, full-cream milk powder and vegetable oil fortified with micronu-trients, provided in a foil-wrapped packet. A packet3 provides500 kcal of energy in each 92g sachet. It is given at dose of175 kcal/kg body weight/day (733 kJ/kg/d) and providesabout 5.3 g protein/kg/d. It was formulated for the manage-ment of severe acute malnutrition. Table 1 shows the maindifferences and advantages of RUTF over the previous prod-ucts, such as F75 and F100, used to treat severely malnour-ished children,. Another RUF called Plumpy’Doz ® is designedas a ‘foodlet’ or supplement to be given as 3 teaspoons a day(approx 45g) and to the recommended intake of micronutri-ents and some essential fatty acids. This is now being widelyused and promoted by agencies such as UNICEF.

Effect of RUTF on gains in weight, height and MUAC

Seven studies were identified which provided standardiseddata on weight gain over the first four weeks of treatmentwith an RUTF and reported data on z-scores of initial anthro-pometric indices, to control for the prior degree of undernu-trition: Amthor et al. (2009)27 who studied 826 children inMalawi, 99% of whom had oedema; Ciliberto et al. (2005)28

who studied 992 children in Malawi; Ciliberto et al. (2006)29

who studied 171 children identified passively and 48 identi-fied actively, in Malawi; Linneman et al. (2007)30 who studied2,131 severely malnourished children and 806 moderatelymalnourished children in Malawi; Manary et al. (2004)31 whostudied 69 children in Malawi (but did not report the SD ofmean weight gain); Ndekha et al. (2005)16 who studied 20 chil-dren in Malawi given RUTF to take home; and Sandige et al.(2004)32 who compared 125 children treated with Plumpy’Nutwith 125 children treated with a locally made RUTF, in Malawi.

Figure 2 shows the relationship between weight gain ing/kg/d and children’s initial anthropometric status beforetreatment with RUTF. Figures 2A and 2B show that an expo-nential fit gives a higher value of r2 than a linear fit, which in-dicates a stronger association. An exponential fit was alsobetter for weight-for-age and weight-for-height, so only theyare shown. An exponential fit is biologically more appropriateas it indicates that the rate of weight gain increases with thedegree of undernutrition, and that there is slower weight gain

Table 1. The major differences between an RUTF, such as Plumpy’Nut, and F100, both used to treat malnourished children.

RUTF F100

How provided Individual 92g sachets Bag of 114g of dry powder

Shelf life at shipment 2 y 1.5 y

Preparation required on site None Mixed with 500 ml warm, clean water

Amount per child/day 1.5 – 2.5 sachets (140 – 230 g) 750 – 1250 ml

Utensils required No Yes

Life when being consumed Several hours <1 hours

How consumed Sucked from sachet Given with spoon or cup

Child needs assistance to consume No Yes

Energy density, raw 5.4 kcal/g 5.3 kcal/g

Energy density as consumed 5.4 kcal/g 1.0 kcal/ml

Cost/kg Sold by UNICEF (2011)

Made in MalawiPlumpy’Nut (Nutriset, France)

USD 3.91USD 2.60USD 4.56

USD 4.22

ResultsThe impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidence

15

even among relatively well nourished children, indicated bythe positive value of the first coefficient (a) in the equation forthe line, y = aebx in which e=2.71828. A linear fit would give anegative weight gain among well nourished children whereasin fact it is just slower than among malnourished children. The average weight gain shown by the treated children inFigure 2 ranges from 2.0 g/kg/d (95% CI 1.35, 2.65) to 5.6g/kg/d (95% CI 4.92, 6.28), which is substantially greater thanexpected from the WHO reference data. An analysis of datafor boys aged 6 - 24 months in the WHO reference sample in-dicate average gains of 0.16 g/kg/d for children with a z-scoreof weight-for-age of -3 and 0.24 g/kg/d for children of medianweight for age. An analysis was done for this review of data reported onthe unstandardised average weight gains of 11,941 childrenat 11 sites in three countries, Ethiopia, Sudan and Malawi in areview of community based therapeutic care.13 The range inweight gain was 3.0 to 6.6 g/kg/day (n=11) with a weightedaverage of 5.45 g/kg/day (95% CI 4.58, 6.31). However these observed weight gains are substantiallylower than theoretical estimates of 10 g/kg/d weight gainbased on an energy intake of 175 kcal/kg/d (733 kJ/kg/d),which is the amount usually given as RUTF.33 A study of 6,363children treated at an in-patient centre in Bangladeshachieved an average of 11 g/kg/d.34 Another study inBangladesh of children also achieved weight gains of 10g/kg/d in the first 8 days of treatment followed by 7 g/kg/d inthe following 10 days.35 The average calculated over 28 days

may have been lower however, which reiterates the point thatsuch data need to be standardised over the same period ofmeasurements, ideally over 28 days, if they are to be com-pared reliably. Figures 2B, 2C and 2D indicate that the rate of weight gainis most strongly predicted by height-for-age (r = 0.826) fol-lowed by weight-for-age (r = 0.782) and then weight-for-height (r = 0.652), which is unexpected, although all arestatistically significant associated. This means that the weightgain of children given RUTF is more strongly associated withthe prior degree of stunted linear growth than the prior de-gree of underweight (which is a composite indicator ofstunted growth and of wasting) or prior wasting alone. Thereasons for this are unclear but may be related to the fact thatthe data on weight gain is standardised by initial body weightand the highest rates of weight gain may be achieved by thelargest children, who are also the tallest. Figure 3 shows that average gain in height in the samestudies is most strongly associated with height-for-age (r =0.572) but is not associated with weight-for-age or weight-for-height. The average gain in MUAC in the same studies wasnot associated with any anthropometric index (data notshown).

The fact that all of the seven studies were done in Malawi isof concern. It would be useful to have data from childrentreated in other countries in Africa as well as in Asia and theIndian sub-continent. Given that RUTF is being widely usedelsewhere, such as in Kenya, Niger, Sudan and Ethiopia, data

A. Height-for-age vs weight gain (linear fit) B. Weight-for-height vs weight gain (exponential fit)

Wei

ght g

ain

(g/k

g/d)

7

6

5

4

3

2

1

0

Wei

ght g

ain

(g/k

g/d)

7

6

5

4

3

2

1

0

Initial z-score of height-for-age0.0 -0.5 -1.0 -1.5 -2.0 -2.5 -3.0 -3.5 -4.0 -4.5


y = -2.2859x - 3.3534R2 = 0.5932

y = 0.4535e -0.6623x

R2 = 0.6829

Figure 2. The observed relationship between weight gain in grams/kilogram initial body weight/day (g/kg/d) and initialz-scores of height-for-age (with linear or exponential fitted lines), weight-for-age and of weight-for-height (both withexponential fits). The estimated 95% CI are also shown except for one point which has no CI for weight gain.

C. Weight-for-age vs weight gain (exponential fit) B. Weight-for-height vs weight gain (exponential fit)

Wei

ght g

ain

(g/k

g/d)

7

6

5

4

3

2

1

0

Wei

ght g

ain

(g/k

g/d)

7

6

5

4

3

2

1

0



y = 0.715e -0.5086x

R2 = 0.6113y = 1.3577e -0.5217x

R2 = 0.425


16

Figure 3. The observed relationship between height gain in mm/d and initial z-scores of height-for-age (with linear or exponential fitted lines), weight-for-age and weight-for-height (with exponential fits). The x and y 95% CI are also shown.

should be obtainable. The data required are listed in the sec-tion on Further Research, below. The study cited above by Ciliberto et al. (2005) comparedRUTF with the then standard treatment for severe under-weight, F75 followed by F100.28 The 992 children treated withRUTF gained 3.5 g/kg/d (95% CI 3.27, 3.73) compared with aweight gain of 2.0 g/kg/d (95% CI 1.01, 2.99) by the 186 chil-dren given the standard treatment. The rate of weight gainof children treated with RUTF was 1.8 times faster than chil-dren given the standard treatment and they also recoveredmore quickly (P < 0.001), defined as a z-score of weight-for-height of >-2.

Analysis of weighted mean averages

An analysis of the average gains in weight drawn from ninesamples of children in six studies of RUTF with a total samplesize of 5,565 children, indicates that the average gain,weighted for the sample size in each study, is 3.6 g/kg/d (95%CI 3.0, 4.2). 32, 28, 36, 29, 30, 27

An analysis of the average gains in height drawn from ninesamples of children in six studies of RUTF with a total samplesize of 5,565 children, indicates that the average gain,weighted for the sample size in each study, is 0.3 mm/d (95%CI 0.25, 0.37). 32, 28, 36, 29, 30, 27

Other studies of weight gain

There have been other studies of RUFs, some in comparisonwith other foods, but the data have not been presented in astandard manner that allows comparisons between studies.

A small trial in Malawi of RUF in comparison with maize-soyflour indicated that the beneficiaries actually received lessthan 50% of the food they were given, so the effectiveness ofthe study was compromised.37 Another study in Malawi com-pared RUTF with a very small group given a corn-soy blendedfood and reported a faster gain in weight (3.1 vs 1.4 g/kg/d, P< 0.001).36 A third trial in Malawi compared the weight gainof moderately malnourished children given either a lipidbased nutrient supplement or CSB with the weight gain of anuntreated control group and reported a small extra weightgain of 150g of borderline statistical significance (P = 0.05)over 12 weeks by the group given the RUTF.38

A comparison of data before and during a civil war inGuinea-Bissau in 1998-99 suggested that a supplementaryfeeding programme prevented the deterioration of nutri-tional status.39

A cluster trial in Haiti examined the effectiveness of givingfood to all children aged 6 – 23 months to try to prevent un-dernutrition compared with giving food only to underweightchildren <5 years of age to help them recuperate.40 Theprevalence of stunting, underweight and wasting were 4% to6% lower in the preventive group than in the recuperativegroup and the mean z-score of weight-for-age and weight-for-height were 0.24 z-scores higher (P <0.001).40

A trial in Malawi compared two fortified spreads, one madewith milk and peanuts, the other made with soya andpeanuts, against a traditional CSB for treating moderately un-derweight children.14 Recovery was quicker in children given

A. Height-for-age vs height gain (linear fit) B. Height-for-age vs height gain (exponential fit)

Hei

ght g

ain

(mm

/d)

0.45

0.40

0.35

0.30

0.25

0.20

0.15

0.10

0.05

0

Initial z-score of weight-for-age0.0 -0.5 -1.0 -1.5 -2.0 -2.5 -3.0 -3.5 -4.0

Initial z-score of weight-for-height

y = -0.0871x + 0.012R2 = 0.2868

y = 0.0925e -0.3512x

R2 = 0.3384

C. Weight-for-height vs height gain (exponential fit) D. Weight-for-height vs height gain (exponential fit)

Hei

ght g

ain

(mm

/d)

0.45

0.40

0.35

0.30

0.25

0.20

0.15

0.10

0.05

00.0 -0.5 -1.0 -1.5 -2.0 -2.5 -3.0 -3.5 -4.0

Hei

ght g

ain

(mm

/d)

0.45

0.40

0.35

0.30

0.25

0.20

0.15

0.10

0.05

0

Initial z-score of weight-for-age0.0 -0.5 -1.0 -1.5 -2.0 -2.5 -3.0 -3.5 -4.0

Initial z-score of weight-for-height

y = 0.2358e-0.0443

R2 = 0.0097

Hei

ght g

ain

(mm

/d)

0.45

0.40

0.35

0.30

0.25

0.20

0.15

0.10

0.05

00.0 -0.5 -1.0 -1.5 -2.0 -2.5 -3.0

y = 0.265e -0.0119x

R2 = 0.0005


17

nutritional rehabilitation unit’ so not all the increase could beattributed to the new diet.43 Other outcomes also improved;the cure rate went up and the death rate and default rate fell.43

The food cost €0.056 a serving including labour and fuel.43 Thelack of standardisation of weight gain as g/kg/d over a periodof 28 days makes it impossible to assess whether the 143% im-provement in weight gain is from a very poor level to a levelthat should normally be achieved, or is a significant develop-ment in treating wasted children. Severely malnourished children in Niger were treated in oneof three situations: for their entire treatment at a therapeuticfeeding centre (TFC) with F100 and RUTF (n=660); at a TFC firstthen at home (n=937); or were exclusively treated at home withfood prepared at home plus RUTF (n=340).44 The data for eachgroup cannot be compared as subjects were not randomly al-located to each group and the children treated at home wereless severely wasted, so there was selection bias. The rates ofweight gain were also not standardised as the mean length ofstay until the children were classified as recovered was 17.4,39.4 and 29.0 days in each of the three groups respectively andthe children gained 20.8, 10.1 and 9.7 g/kg/d.44

Effects of antibiotics

Malnourished children in a camp for internally displaced peo-ple in south Sudan were randomly allocated either to receivetreatment with injected ceftriaxaone for two days or amoxi-cillin given orally for five days plus the same standard thera-peutic nutritional treatment with F100 followed by RUTF plusmicronutrients and deworming. There was no statistically sig-nificant difference between groups in terms of the percentagewho had a weight gain of ≥10 g/kg/day over the first 14 daysor in terms of the recovery rate or case fatality rate.45 The treat-ments were concluded to be similar, so the costs are likely tobe the main distinguishing feature.

Coverage surveys or programme evaluations

Two simultaneous surveys in Malawi estimated the periodand point coverage of TFCs or community-based therapeuticcare (CTC) in two districts using a centric systematic area sam-pling method.46 The CTC programme gave higher period cov-erage (73.6% vs 23.6%) and higher point coverage (60.0% vs20%) of children <5 years old whose weight-for-height was≤70% of the NCHS median.46

An evaluation was undertaken in Burundi of the impactand ‘appropriateness’ (not defined) of 20 TFCs and 224 sup-plementary feeding centres, one in every province of thecountry, during a humanitarian emergency between 1999and 2004.47 The evaluation claimed that the nutrition pro-gramme was responsible for a difference in death rate esti-mated by unpublished surveys of children aged <5 years of6/10,000 in 2000 compared with 3.1 to 4.9/10,000 in 2004, butno confidence intervals around these estimates were given,47

so no evidence for a significant difference is provided The im-pact on nutritional status was said to be ‘limited’ as the preva-lence of acute malnutrition (not defined) was 8% in 2000 and7% in 2004.47 The prevalence of severe malnutrition was saidto have worsened from 0.5% to 1.1%. Although no confi-dence intervals are given, it would be necessary to have asample of 15,000 children in each group to be able to detect

the two fortified spreads, but of the 1,362 children treated atotal of 109 or 8% in each of the three groups developedoedema while being fed.14 Why children should develop kwa-shiorkor while being treated for malnutrition is unknown. Thefactors associated with developing oedema were beingyoung, being wasted and not gaining weight in the first twoweeks of treatment, although none of these factors was spec-ified quantitatively to suggest risk factors that could be usedto screen or identify potentially susceptible children.14 The au-thors called this ‘intriguing’ rather than worrying. It suggeststhat there may be risks in giving young, moderately wastedchildren an energy- and nutrient-dense food. A blanket supplementary feeding programme in Niger pro-vided children aged 6-26 months who had a MUAC >110 mmwith 4 x 325g pots of Plumpy’Doz®, which is designed to begiven as a daily dose of three tablespoons or about 50g/d, andshould last about a month.41 The authors claimed that fewerchildren were given therapeutic feeding than in previous years,and that the proportion of children requiring intensive carewas lower than in children not registered for the programme.The results could be explained by a concurrent improvementin food security and selection bias in identifying beneficiariesof the programme, as there was no random allocation. Also in Niger, 12 villages were randomised into two groups:in one group all non-malnourished children aged 6-60months were given a daily packet of RUTF while nothing wasgiven in the control villages.42 There were significant differ-ences in changes in z-scores of weight-for-height after 8months and the intervention was estimated to have led to a36% difference in the incidence of wasting and a 58% differ-ence in the incidence of severe wasting.42 However there wasa higher prevalence of stunted or severely stunted children inthe control villages at the start of the study, which Figure 2Dshows to be associated with subsequent weight gain. The au-thors claimed that the difference was a ‘reduction’, but it couldhave been due to an increase in the incidence of wasting inthe control villages and no change in the intervention villages:see Figure 1D, rather than Figure 1A. A study in Niger compared RUTF with CSB as a treatmentfor moderately malnourished young children.15 The criterionfor recovery was achieving and sustaining a weight-for-heightof 85% of the median for 2 weeks. Children given RUTFgained 1.1 g/kg/d more than children given CSB until recov-ery, so recovered more quickly.15

An analysis of the weight gain of Ugandan children in a nu-trition rehabilitation unit reported their mean absolute weightgain which was corrected statistically for ‘days in care’ (presum-ably days in the unit) and age at admission (rather than weightor degree of undernutrition, which might have been better).43

The children were sampled from all admissions at the sametime of year over three successive years. In the first year chil-dren were given milk alone (F75 followed by F100) and thenin the next two years a semi-solid porridge with an energydensity of 4.4 kcal/g. The authors reported a mean weight gainof 525g (95% CI 373,676) in 2001, 771g (95% CI 618,925) in2002 and 1,274g (95% CI 1,120, 1,427) in 2003.43 This occurredat the same time as ‘various improvements were made in the


18

a statistically significant difference between 8% and 7%, solimited is an inappropriate term; there was probably no dif-ference. The fact that only 8% of children were acutely mal-nourished, presumably wasted, suggests that malnutritionwas not a major problem either, as this is prevalence is onlyclassified as poor by the WHO.1 The evaluation also examinedperformance indicators. Surveys in Ethiopia, Malawi, Niger and north and southSudan of 12 community-based therapeutic care programmeswere undertaken to identify why the caretakers did not enroltheir malnourished child in the programmes.48 The top fivereasons given were previous rejection (38.6%), condition notrecognised as malnutrition (18.8%), no belief in the pro-gramme (14.0%), the child had relapsed (11.3%), and the dis-tance to the site of the programme (10.8%).48

Conclusions

The development of RUF seems to have improved the cover-age of treating malnourished children with moderate and se-vere wasting in all circumstances, both in emergencies and insituations of chronic undernutrition, mainly, it seems, becauseit does not need to be prepared and can be taken home. Thedevelopment of RUTF has extended nutritional therapy intocommunities and away from residential treatment centreswhich have disadvantages for all but the sickest, most mal-nourished children who need medical treatment. As RUTF re-quires no preparation, it has a long shelf life and as many, butnot all, malnourished children are old enough to feed them-selves, this frees the mother from the responsibility of stayingwith a malnourished child at a therapeutic feeding centre.The child can be treated at home, away from the danger ofnosocomial infection as well. The risk is that the supply ofRUTF may be shared or even sold, and that it removes fromthe mother the responsibility and need to prepare nutritiousfood for her malnourished child. As this review has focused on papers published since 2004,most of them have studied the rate of weight gain of childrengiven RUTF, and most of the studies have been done inMalawi. There is a need for studies in Asia and the Indian sub-continent to see if the same effectiveness can be achieved ei-ther with an RUTF based on peanuts or one based on anindigenous source of protein, because as there is some resist-ance to giving children a food that is not commonly eaten bytheir parents. It seems unlikely that very young children willhave developed any preference and although there may bebenefits of using a locally available source of protein, there isno reason why children in the Indian sub-continent shouldnot be able to digest, absorb and assimilate the nutrients inpeanuts. An interesting study would be to compare a peanut-based RUTF with a gram- (chick-pea) or lentil-based RUTF for-tified with oil to make them equivalent in energy and proteinin both Malawi and India, perhaps. It would also be useful to undertake a general review of therates of weight gain, ideally standardised over the first 4 weeksafter the start of treatment. Although RUTF is rich in fat, proteinand micronutrients, the rates of weight gain are rarely muchmore than 6 g/kg/d and are related in general to the prior de-gree of undernutrition (see Figure 2), but most strongly with

the degree of stunting (see Figure 2B), which seems counter-intuitive. Why is the rate of weight gain not more strongly as-sociated with the prior degree of wasting (Figure 2D)? But 6g/kg/d is less than the old Sphere standard of 8 g/kg/d, whichseems to have disappeared from the recommendations, and isless than the gains of 10 g/kg/d which have been achieved inthe past when treating malnourished children in Bangladesh,although such rates are not standardised for the period of treat-ment. It would be useful to assess whether RUTF, for all its easeof use, is a less effective treatment for wasting than other foods.Such an analysis is beyond the scope of this review but is pro-posed below in the section Research questions.

6.3 Micronutrients distribution

Surveys of micronutrient deficiencies in people living inrefugee camps have shown that they are common in all agegroups.49 However to be able to interpret the data it is neces-sary to know the prevalence of deficiencies in the populationfrom which the refugees originally came; if they are the same,then being a refugee is no greater risk. The prevalence of micronutrient deficiencies may even belower among refugees, because they tend to be given forti-fied foods. A study of the concentration of iodine in the urineof adolescents in six refugee camps in five African countriesshowed evidence of excessive iodine consumption while theprevalence of visible goitre was as high as 7%, probably dueto hyperthyroidism.50 It seems reasonable to conclude that asmost solid food given to refugees is fortified with iodine thereis a risk of excessive consumption, especially if they are givenover-iodised salt as well.50

A randomised, controlled trial in Algeria studied thegrowth of children given one of five treatments, a highly for-tified snack plus metronidazole (a broad spectrum antimicro-bial), a highly fortified snack without metronidazole, anunfortified snack with metronidazole, an unfortified snackwithout metronidazole; and a control group given nothinguntil the study had finished.51 The fortified snack consisted of50g of a paste of peanuts, milk power and oil that providedabout 320 kcal/d and was fortified with between 1.5 and 3.0times the reference intakes of 11 vitamins and seven minerals.As there was no evidence of an effect of metronidazole, thetwo pairs of groups were merged. The children given the for-tified snack gained 9 ± 3 mm/month in length in the first threemonths of treatment compared with 7 ± 3 mm/month in thegroup given the unfortified snack and 6 ± 3 mm/month in thecontrol group, which was said to be statistically significant (P<0.001).51 If the ± values are presumed to be standard devia-tions around a normal distribution, then 95% CI can be esti-mated as follows: fortified 9 mm 95% CI 8.43, 9.57; unfortified7 mm 95% CI 6.42, 7.58; control 6 mm 95% CI 5.12, 6.88. Thisindicates statistically significant differences between thegroup given the fortified spread and the other two groups,but not between the unfortified spread and the control group.There were also statistically significant differences in the in-crease in haemoglobin concentration over a six month periodof supplementation: fortified supplement group, 37 g/L (95%CI 29.4, 44.6); unfortified supplement group 19 g/L (95% CI16.1, 21.9); and control group 16 g/L (95% CI 11.0, 21.0).51 The


19

prevalence of anaemia fell by 90% in the group given the for-tified supplement, by 40% in the unfortified supplementgroup and by 27% in the control group. The haemoglobinconcentration of 40 children in the untreated control groupincreased by 17.6% from 93 g/L (95% CI 85.9, 100.1) to 109 g/L(95% CI 103.1, 114.9) over a period of 6 months without anytreatment, which was said not to be statistically significant (p978)51 although the estimated 95% CI suggest otherwise. Thestudy concluded that the differences were due to the mi-cronutrients. Without the control group, the study wouldhave greatly over-estimated the attributable effect of the sup-plements, particularly on the haemoglobin concentration. An evaluation of a food for work programme at five differ-ent sites in Indonesia attempted to assess if the programmehad an effect on the odds that children or mothers wereanaemic in comparison with control households not takingpart in the programme.52 However the ‘control’ householdswere significantly different from households eligible for theprogramme at the start of the study in terms of education, in-come and the prevalence of stunted children,52 so there wasstrong evidence of selection bias. The analysis reported sta-tistically significant odds of being anaemic between mothersin the two treatment and control groups in one of the fivesites and no statistically significant odds between children inthe two groups in the five sites52. The probability of finding astatistically significant difference in 10 statistical tests bychance alone is P = 0.1 which is only twice the usual thresholdfor assessing statistical significance of 0.05. This illustrates theneed to be able to allocate subjects randomly to interventionand control groups from the same pool of eligible subjects,either individually or in clusters so that the only difference be-tween subjects is the intervention, not education, income orany other factor that could influence the outcome. An attempt was made in Guinea-Bissau to assess whethervitamin A supplements had an effect on death rates of youngchildren during a civil war.53 It was estimated that the risk ofdying increased by 89% during the war. Two methods wereused: a variation in the delivery of vitamin A described as astepped wedge design, although there was no random allo-cation, and a comparison of death rates before the war, whenvitamin A was not provided, with death rates during the war.There was no statistically significant difference in the deathrate estimated in March 1999 between children given vitaminA supplements in October 1998 and eligible children who didnot get treatment, but the death rate among children givenvitamin A during the war was 12% lower than before the war(mortality ratio 0.88, 95% CI 0.41, 1.87).53

Ten supplementary feeding sites in Haiti were randomisedinto two groups: one group of children (n = 254) at five siteswere given micronutrient sprinkles for 2 months and theother group (n = 161) at five sites acted as an untreated con-trol.54 Anaemia was defined as a haemoglobin concentration<100 g/L, which is not the standard WHO definition.55 Al-though there was a significant difference in the prevalence ofanaemia between the treatment and control groups beforethe treatment started (52.3% vs 36.6%), a likely consequenceof the small number of clusters, the haemoglobin concentra-tion increased over a 2 month period by an estimated 5.5 g/L

in the treated group compared with -1.0 g/L in the untreatedcontrol group (both means were adjusted for initial haemo-globin concentration, age and sex).54 The prevalence ofanaemia fell to 28.3% in the treated group and increased to44.8% in the untreated group.54

A study in a refugee camp in Zambia estimated the haemo-globin concentration of young children, adolescents andwomen before and after their maize meal rations were fortifiedwith seven vitamins, iron and zinc.56 There was no statisticallysignificant change in the haemoglobin concentration ofwomen and adolescents or in the prevalence of anaemia overa period of 12 months, although the prevalence of anaemiaactually increased in both groups. The mean haemoglobinconcentration of young children increased by 8.7 g/L whichwas adjusted for an estimated increase of about 2.7 g/L due toage. The prevalence of anaemia fell by 23.4%, from 47.7% to24.3% (P <0.001).56 There were significant improvements inthe concentration of serum transferrin receptor and serumretinol concentration.56 However the lack of an untreated con-trol group and the lack of a significant effect on the haemo-globin concentration of adolescents and women makes itdifficult to assess the effectiveness of the intervention as therewas not a consistent effect among all three study groups. A weak evaluation was done in Palestine by comparingtwo random samples of children, one selected in September2004 before giving children iron supplements, rations of food,health education and free medical treatment, and then an-other random sample eight months later, in May 2005. 57 Theauthors claim that the intervention led to a 50% drop in theprevalence of anaemia from 30.1% to 18.8% (which is only a38% difference), and a 70% drop in the prevalence of wastingfrom 6.0% to 1.4% (which is a 77% difference).57 The lack of acontrol group to account for seasonal changes in food pricesand disease transmission, presumably during the winter,means that it is not possible to estimate the attributablechange due to the package of interventions. It would alsohave been better to have studied a cohort of children, be-cause it is possible that the two random samples were differ-ent. No data were presented on socio-economiccharacteristics of the two groups of children to indicate thatthey were from similar households. A study in two refugee camps in Tanzania compared theconcentration of haemoglobin and serum transferrin receptorin children in households in one camp that were all givenstainless steel cooking pots with children in another campthat used clay or aluminium cooking pots.58 There was nochange in haemoglobin concentration over a period of a yearin either group and there was no difference between studygroups. The concentration of serum transferrin receptor washigher in the group using steel cooking pots, so their iron sta-tus was worse that the group using clay or aluminium pots.58

The lack of effect could have been because the iron in stain-less steel is poorly soluble compared with cast iron cookingpots which have been shown in studies in Ethiopia59 andBrazil60 to lead to an improved haemoglobin concentration.The subjects of the study did not like cast iron cooking post,so steel was used instead but it seems to have rendered theintervention ineffective.


20

Conclusions

The few studies described here are generally weak in their de-sign: they compare subjects before and after an intervention,suffer from selection bias or have non-randomly selected ‘con-trols’ so cannot estimate the attributable change in micronu-trient status. One of the main problems is that thehaemoglobin concentration of children tends to increase withage, so this needs to be controlled for in any study, ideally byhaving an untreated control group. The only study with sucha control was done in Algeria and showed an 18% increase inthe haemoglobin concentration of the control group over aperiod of three months, which suggests that other factors hadan influence, not just age. There are also inconsistent resultsin some studies of multiple groups: why in the refugee campin Zambia was there no apparent effect of fortifying maizeflour with micronutrients on the haemoglobin concentrationof adolescent children and adult women, but there was an ef-fect on young children? Again, an untreated control groupcould have told us whether the increase in the mean haemo-globin concentration of young children was due to the forti-fied food or occurred naturally in whole or in part due tonatural processes associated with age. It is very difficult to do properly controlled studies in emer-gencies let alone in normal circumstances among undernour-ished populations, and it is not clear that such studies arenecessary, either. If micronutrients are efficacious during anon-emergency and have a measureable biomedical effecton micronutrient status, then there is no reason why that ef-fect is not likely to occur during an emergency. Surely theissue during an emergency is delivering micronutrients eitherin food or as supplements, and achieving high coverage tominimise the risk of deficiency. As most foods given to peopleduring emergencies are now fortified with micronutrients, itis possible that intakes are better than normal for some peo-ple. The fact that people in several long-term refugee campsin Africa show signs of an excessive intake of iodine50 is per-haps an indicator, although iodine is obtained from iodisedsalt as well as rations of food, so there is more than one source.Nevertheless, humanitarian agencies are aware that the mi-cronutrient status of people during emergencies needs to besustained and there has been a joint WHO/WFP/UNICEF state-ment on the topic.61

6.4 Food security or agricultural interventions

No papers were found with any such interventions during anemergency. A study in Bangladesh compared the production and con-sumption of vegetables by households that had participatedin a homestead gardening programme and ‘control’ house-holds that had not.62 The design was weak because the con-trols were households in the same villages that had notparticipated or had not been selected for the programme forsome reason, and were more likely to have no regular income,so there was selection bias. Unsurprisingly households thathad participated in the programme grew, consumed and soldmore vegetables than households that had not. The medianweight of vegetables grown was 135kg vs 46kg and the me-dian weight of vegetables consumed was 85kg vs 38kg.

About two thirds of the participant households generated amedian of Taka 347 (approx USD6.8 in 2005 prices) and spentmuch of it on buying extra food.

6.5 General ration distribution programmes e.g. food-for-work

The fact that rations are given to the general populationmeans that estimating the impact on any outcome is very dif-ficult. A study of death rates in areas of Ethiopia that were ei-ther affected or unaffected by the drought of 2002-03concluded that receiving food aid had a small but significanteffect on reducing death rates among children, but that theoverall lack of an effect on child mortality was perhaps moreremarkable.63

In Bangladesh, women and children in households thatwere randomly selected for a cash-for-work programme in thechars (riverine islands and sandbanks) were compared withwomen and children in control households that were not se-lected for the programme. The two groups of householdswere stated to be similar in terms of the adults’ occupations,the number of household members and the age of the adultfemale.64 There were no significant differences between thegroups in terms of the anthropometric status of children orwomen.64 There were statistically significant differences in themean increase in weight and height of children and the meanweight and MUAC of women also differed.64 However it is notclear how beneficiaries were chosen or not chosen for theprogramme in the first place; if it was not random then therecould have been selection bias. The targets of such pro-grammes are usually the most needy, and some method ofidentifying them is usually applied. Random selection of ben-eficiaries to the programme or to a control group would havehelped to ensure that there were no differences between thestudy group in some unmeasured factors that could have ledto the differences, rather than the intervention. Conclusions

The main problem with assessing the impact of general ra-tions or food for work is that they are either general, so every-one is eligible and no control group can be achieved, or thetargets are the most needy and are most likely to have under-nourished children. So in both instances having an untreatedcontrol group during a humanitarian emergency is unethical. However here is potential to assess the impact of pro-grammes during non-emergencies using a stepped wedgedesign. This design utilises the fact that programmes canrarely be implemented immediately to reach the whole targetpopulation, so they are staggered or phased in. If the units ofdelivery such as villages within a single area could be ran-domly assigned to receive the intervention at different inter-vals, then there is the potential for people in later villages toact as controls for the first phase villages. If the villages couldbe involved in the process of randomisation and understoodthe reasons for it, then it may be acceptable and help to min-imise the risk of contamination. It is somewhat surprising thatan agency such as the World Food Programme (WFP) has notattempted such a design to evaluate the nutritional impactof food for work programmes.


21

6.6 Conditional cash transfers, loans or vouchers

There have been few studies of such intervention during anemergency, perhaps because the ability to purchase neces-sary food and goods are often greatly reduced or becauseprices are inflated. During emergencies, people at risk tendto be given food or gifts in kind to replace items lost or sold. Surveys during a drought in Ethiopia of households thatwere clients of a microfinance programme found that the chil-dren of female clients were less likely to be malnourished thancommunity controls, and that male clients and communitycontrols were more likely to have received food aid.56 How-ever it is possible that clients were systematically differentfrom community controls, which was indicated by the mem-bership of a microfinance programme. Random allocation ofpotential subjects to the programme and to a control group,so that confounding variables were equally distributed be-tween groups before the intervention, is necessary to showthat being a microfinance client has an effect on any outcome. During a conditional cash transfer programme in Brazil,young children in beneficiary households were comparedwith children in households that had been selected for theprogramme but were excluded as a result of ‘quasi-randomadministrative errors’. For example, potential beneficiarieswere mistakenly rejected by a computer programme becausetheir name included a non standard character such as é, ô orç.65 Data on anthropometric measurements were obtainedfrom growth monitoring cards. Six months after the start ofthe programme, the children of beneficiaries had a weight-for-age that was 0.13 z-scores lower (worse) than non-bene-ficiaries.65 There was also an inverse dose-responserelationship: each additional month in the programme wasassociated with a rate of weight gain in the interventiongroup that was 31g lower (worse) than in the control groupand amounted to about 180g over six months.65 This was at-tributed to the fact that mothers may have thought that theirparticipation was dependent on their child being under-weight and that their benefits would be lost if the childstarted to grow well.65 This could be a harmful effect of growthmonitoring linked to benefits. A conditional cash transfer programme was implementedin 6,400 poor communities in Mexico.66 The cash amounted amaximum of $90 for a household containing primary schoolchildren and $160 for a household containing secondaryschool children and was conditional on children’s enrolmentin school and parents attendance at health education ses-sions.66 A total of 506 poor communities were randomly sam-pled in proportion to size and 306 were randomly assignedto start receiving the cash transfers in April 1998 while the re-maining 200 were started 20 months later, in December1999.66 This is a good stepped wedge design. The impact onreproductive health outcomes was assessed in a sample ofwomen in each group of communities who had given birthto a single child, 606 and 419 children respectively in the in-tervention and control communities. After adjusting for alarge number of individual, household and community char-acteristics, a small difference of 127 g (P = 0.02) in birth weightwas found and a 4.6% lower risk of having a low birth weightbaby (P = 0.05).66 It is odd that it took about 8 years for the

data to be analysed and published given that the study hadsuch a good design, perhaps because the differences werenot large and required statistical manipulation. A meta-analysis of the effect of conditional cash transferson health outcome was done under the auspices of theCochrane Collaboration.67 Ten papers reporting data from sixstudies were analysed. The analysis showed evidence of ef-fects on nutritional status and health, but it was hard to at-tribute the effects specifically to the cash transfers as othercomponents may have contributed. Positive effects on an-thropometric status were found in three studies, two clusterrandomised controlled trials (RCTs) and one controlled beforeand after study, but one study (reported above) found a neg-ative effect. No estimates were made of the pooled effect,probably because the number of studies with the same meas-ured outcome was too small.

Conclusions

Cash tends to be an uncommon intervention during an emer-gency unless the commercial infrastructure and supply linesare still operating. A disaster such as the tsunami that struckSri Lanka in 2004 is an example: the damage to housing andinfrastructure was restricted to a narrow band around thesouthern and eastern coast so that supplies of food and goodswere sustained. Again there is considerable potential to collectbetter data in non-emergency situations, perhaps during clus-ter randomised trials or using a stepped wedge design.

6.7 Costs and cost-efficiency

The effects of buying cattle during a drought, called destock-ing, were studied in southern Ethiopia in 2006. Householdsreceived an average of USD 186 from the sale of cattle thatmight have died during the drought which provided 54% ofhousehold income and was used to buy food and health carerather than waiting for food aid.68 The cost-benefit ratio wasestimated to be 41:1 and destocking was given the highestscore of all possible interventions. An analysis of school feeding programmes operated by theWFP in 42 countries in 2005 for about 8.6 million children showedthat cost from $11 to $52 per child per year, with an average of$21.59.69 The types of programme included cooked meals, forti-fied baked biscuits or take-home rations. The main costs weredue to delivery but did not include any financial or economiccosts for schools. Fortified biscuits were considered to offer thebest option, especially in contexts where security was poor. Conclusions

The data on costs and cost-effectiveness of any interventionduring an emergency are generally lacking. This is a little sur-prising given that agencies operating during emergencies areoften given funds specifically for use in that situation so theyknow how much they have spent and, at minimum, record thenumber of people they have helped, so can estimate the costper person treated. Give that non-governmental organisationsroutinely collect lots of data that could contribute to creatingcost menus, such as staff time, mileage of vehicles, tons ofRUTF purchased, and they avidly record the number of bene-ficiaries to show the scale of their programmes, such datashould be relatively easy to extract and analyse. So why not?

22

This update of the review by Duffield and colleagues4

has found few studies published since 2004 on theimpact of humanitarian interventions on thenutritional status of beneficiaries and target

populations. The first review found a similar lack of literaturepublished before 2004, so this is not new, if disappointing.This update has also taken a different approach to presentingthe data and to discussing it, largely so as not to repeat toomuch of the background given in the first report, but also totry to make points in a new, or at least a different, way. A total of 59 papers have been reviewed in some detailhere, several of which were not about emergencies but wereincluded here because there were so few papers in general,let alone in emergencies. Other papers could have beenreviewed, so the distinction is more arbitrary than methodical.There was not time to review several papers in the field ofinterventions that could improve nutritional status, such asmicronutrients sprinkles or deworming, an intervention thatis often applied during an emergency as a matter of routine,but not specifically because of the emergency. The nutritionalbenefits of deworming have been reviewed extensivelyelsewhere.70 There are many other interventions that can beprovided during an emergency, but most are deliveredcollectively or as a package: permutations of food in variousforms or delivery mechanisms, money to buy food or helprestore livelihoods, vaccinations, micronutrients, deworming,

bed nets, clean water, sanitation, health education,interventions for feeding young children etc. The list ofoptions is typical of humanitarian aid in an emergency. It is very likely that there are a large number ofunpublished studies, evaluations and analyses in the so called‘grey literature’, mainly produced by humanitarian agenciesas reports of internal evaluations or to describe the impact ofprogrammes to donors. While some of this work may bereported within non-peer reviewed publications such as FieldExchange (www.ennonline.net/fex) or in publications such asthe monthly in-house journal of Médecins sans Frontières ormay be presented at international technical meetings andthen appear as a brief abstract in the proceedings,71-75 muchof it will be generally unavailable. There are many reasons forthe lack of published literature, which include the difficulty ofconducting research in emergencies, a lack of technicalcapacity to design research studies that would be acceptablefor publication in peer reviewed journals, and lack of time,resources and skills to complete the lengthy process ofwriting, reviewing and revising a paper that is necessary topublish in a peer reviewed journal. Because of the lack ofevidence, this means that programme decisions have oftenbeen based on extrapolation, anecdotes and intuition, withhopes of doing the right thing.76 A proposal to extract thebest data from the grey literature and examine its strengthsand weaknesses is outlined below, in the Recommendations.

Discussion7

The impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidenceDiscussion

Mesfin Teklu, WV, Ethiopia, 2006

DiscussionThe impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidence

23

Ready-to-use therapeutic foodsThe most significant advance in the emergency nutritionsector over the past 30 years has been the treatment of severewasting in the community using RUTFs, mainly because of thestrong evidence for the benefits of the therapeutic food itselfand for allowing mothers of malnourished children to take ithome so that their children are treated in the community, notin a treatment centre or hospital. This evidence is providedby a number of robust, well-designed studies undertakensince 2000, which were steered by a consortium of donors,implementing agencies and research groups. The researchshowed that a community-based approach of acutemalnutrition management using RUTF was as effective astreating children in hospitals or feeding centres, and achievedfar greater coverage. There is need for similarly robustevidence for other interventions currently employed innutrition emergencies. Without evidence, donors andimplementing partners cannot be sure that their currentprogrammes are adequately meeting the needs of vulnerablepopulations in nutrition emergencies, or may even be doingharm as a couple of studies reported here have shown. Thisis discussed further, below, in the section Research questions. The main body of published work since 2004 relates tostudies on the performance and impact of community basedtreatment of wasting using RUFs, and mainly RUTF. Thesestudies most often compare different types of food ordifferent delivery mechanisms, so are effectively comparativeclinical trials. Not all these studies could be included in someof the analysis presented in Figures 2 and 3 because themethods were not the same, such as the period of feeding, orthe data were not reported in a standardised way. The lack ofuntreated control groups in many of these studies isunderstandable, but this prevents accurate estimates of thebenefit attributable to the treatment. Problems with methodswere also evident in studies of general food and cash transferprogrammes, blanket feeding programmes, and cost-effectiveness studies of community treatments of wasting.This is discussed further, below, in the section Researchmethods. Very few of the studies of RUTF have had an untreatedcontrol group unless the study has been done using clusters;if individuals had been randomised some would get RUTF andothers would not, so it is difficult to find willing participantsfor such a study. The cluster trial in Niger with untreatedcontrols was probably undermined by major differences inthe initial prevalence of stunting between the two groups ofvillages – an almost 10% difference.42 But it can difficult toachieve similar group means when a large sample size isdrawn from a small number of clusters, in this case 1000children in six clusters per group, or 167 subjects per group.This means that children in one outlier village can have amajor influence on the group mean and it is impossible todistribute those outliers between the two groups bystratification as they are aggregated as a cluster. It is betterto have 1000 children in >25 villages per group than 1000children in only 6 villages. Many investigators know to applyan arbitrary design effect of two when doing a sample sizecalculation to study subjects in clusters, but often fail to have

sufficient clusters or to take the clustering of subjects intoaccount when estimating confidence intervals for their data. The most useful studies of the effect of RUTF on growthand weight gain have measured the change inanthropometric measurements over a period of 28 days forweight and 56 days for height or length, and have expressedthe gain as grams per kilogram body per day (g/kg/d) forweight gain and in millimetres per day for height gain(mm/d). The studies have also reported the initial z-scores ofweight-for-height, height-for-age and weight-for-age; infuture it would be useful to have MUAC-for-age and BMI-for-age, both of which are calculated by the new WHO Anthrosoftware. As there are several suitable studies of RUTF, itwould be interesting to review past studies to see how theaverage weight gain compared with other older, therapeuticdiets used in different circumstances, if the data could becompared. The two studies in Bangladesh that reported gainsin weight of around 10 g/kg/d, which is considerably higherthan the average of 3.6 g/kg/d estimated from six studies ofRUTF, did not report weight gain averaged over a period of 28days. The analysis for this review of data from the fewstandardised studies of RUTF offered interesting insights intothe relationship between initial anthropometric status andthe rate of weight gain, but the review of the literature on thistopic was not exhaustive, so further research is proposed (seeFurther research, below).

Research questionsThis review has shown that the type of research currentlybeing published fails to answer adequately many of thequestions posed by the studies. Furthermore, there are manyresearch questions relating to operational practice innutritional emergencies that are not being addressed. This issometimes due to the practical difficulties of doing researchin emergencies but there are also problems with study design. Some key questions relate to the impact and relative cost-effectiveness of different programme designs. For examplecan cash interventions reduce the prevalence of wasting and,if so, in which contexts and what are key design feature ofthese programmes which will promote a nutritional impact.And what is the cost-effectiveness of blanket supplementaryfeeding compared with an expanded general distribution offood, or how does decentralised, targeted supplementaryfeeding compare with blanked supplementary feeding? What is needed is agreement and prioritisation amongsthumanitarian agencies as to what key questions need to beanswered and consensus about the methods to be used toanswer these questions (see below).

Research methodsThe studies described in this review often have weak methodsand use an inconsistent or un-standardised approach toassessing or reporting impact which contributes to weakconclusions. In many cases, the limitations of the methodsand analysis are poorly described. More standardisation of

The impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidence Discussion

24

methods is needed if coherent answers are to be provided.Currently agencies that conduct research in emergencies(especially impact studies) use their own research tools andmethods, which mean that it is difficult for decision makersto compare or merge findings for meta-analysis. Donors inparticular are increasingly demanding standardisation ofmonitoring and impact assessment. For example, ECHO(European Commission Humanitarian Aid Department) isplanning to adopt minimum reporting standards foremergency supplementary feeding programmes. Whilestandardised approaches must take account of the realities ofthe research environment, such as funding opportunities andthe methods that can be used within an emergency context,broad agreement on the key research questions andapproaches will help guide the type of institutionalarrangements to facilitate research in emergencies.Institutional issues that need to be addressed include thecurrent short term nature of funding, a lack of flexibility withregard to budget lines for research in emergencies, theslowness of obtaining ethical approval for research, flexibilityin programme design to allow research to be done that isadapted to local circumstances, and the lack of coordinationbetween stake-holders such as donors, implementingagencies and research groups, This last issue is beingaddressed by the research consortium model currently beingused by the UK Department for International Development(DFID). This consortium allows for more coordinationbetween bodies funding the research, study designers andthe people who are implementing programmes that will actas vehicles for the research. Another critical challenge for research in emergencies,especially impact studies, is that it is very difficult to have avalid control group to estimate the effect attributable to theintervention alone. This is largely for ethical reasons. Acomparison of two interventions or of the same interventionsdelivered in different ways simply allows an estimate of thedifference in their effects. Control groups of subjects living innon-intervention areas or in households that don’t participatein the intervention are likely to be systematically differentfrom beneficiaries, which means that estimates of attributablebenefit will be wrong. Ideally, both groups should be drawnrandomly from all potential beneficiaries, but this is notalways possible. These sampling issues are rarely addresseddirectly. There are ways to improve methods used in studies duringemergencies that can be applied within the existing constraintsoutlined above. For example, step wedged designs, nestedcase-control studies, dose-response analysis, and comparingcurrent with historical data. If several different lines of evidencecan be obtained, conclusions can perhaps be drawn with adegree of plausibility rather than obtaining a single but strongprobabilistic estimate of statistical significance.

FundingEmergencies provide a very difficult environment withinwhich to conduct rigorous impact studies. Many occursuddenly, so it is difficult to plan. This is compounded by thenature of funding opportunities for research in emergencies.Most funding windows exist within the humanitarian sections

of donor agencies such as ECHO, DFID, and the CanadianInternational Development Agency (CIDA), and tend to beshort-term (rarely more than two years, although somedonors may allow for a one year no cost-extension). It is notunusual for donors to grant funding one or two monthsbefore the grant start date. Research groups are then facedwith having to turn a project proposal into a researchprotocol, recruit researchers, liaise with implementingagencies in order to identify and secure a research site, seekethical approval for the research both in the country of theresearch and the country of the investigators, seekadministrative approval in the country of research, and thenbegin doing the study, by which time the emergency may beover and the main period of impact may have passed. Thisprocess can also easily take 6 months or more of a two-yeargrant. There have been exceptions to this short-term type offunding opportunity, such as the long-term research fundingmade available to Valid International by Concern Worldwideand DFID, which allowed a series of studies to be done thatprovided evidence for the effectiveness of treating acutemalnutrition in communities using RUTF.


25

Recommendations8

As there is likely to be a significant body ofunpublished reports of studies held byhumanitarian agencies which would be useful toappraise and add to the published literature

summarised here, a focused approach is proposed to obtainand summarise the data. The lead nutritionist in each of themain humanitarian agencies could be approached and askedto identify the five best studies or evaluations that they knowof done by their agency in the major fields of humanitarianaid. The reports could then be provided or obtained andcritically reviewed, possibly with some re-analysis of data, ifappropriate and if available. The findings of this review couldthen be presented at an international technical meeting withfour principle objectives:1. Present key findings and conclusions from the selected studies2. Examine the strengths, weaknesses and constraints of the studies with a view to identifying better methods that could be used in emergencies that could lead to stronger or more robust conclusions3. Identify how such reports could be published in future4. Identify further research questions arising from this work.

An agency such as 3ie (www.3ieimpact.org) could be involvedwith this review.

Further researchThe review of published papers and the analysis of extracteddata suggest a number of areas where further research maybe useful: • Rates of weight gain and standardisation of reporting. It would be interesting to search the literature for studies that have reported the weight gains of children given therapeutic foods other than an RUTF based on peanuts. Although this review has estimated that RUTF can achieve an average gain in weight of 3.6 g/kg/d, there is little evidence yet of the relative effectiveness of this form of supplementary food compared with other foods. In order that data can be reliably compared and evaluated in the ways shown in Figures 2 and 3, the following data need to be provided: • the sample size by study group and the proportion of females in each group• the initial mean and standard deviation of the age of each study group in months• the mean and standard deviation or 95% confidence intervals around the change in body weight over 28 days feeding, expressed in g/kg body weight/day• the mean and standard deviation or 95% confidence intervals around the change in height or length over 56 days expressed as mm/day; • the mean and standard deviation or 95% confidence intervals around the change in MUAC over 56 days expressed as mm/day

Recommendations

Indrias Getachew, UNICEF Ethiopia, 2009.

The impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidence Recommendations

26

• the initial mean and standard deviation or 95% confidence intervals around initial z-scores of height-for-age, weight- for-age, weight-for-height. These data would allow comparisons to be made of therates of weight gain between different foods. It would beuseful to have these recommendations critically reviewed andthen widely promoted so that future studies could report datain a standardised way.

• Relapse rates and long term development. The focus on treating acute wasting in the community has been on the effectiveness of RUTF: the mother is given a large supply of foil wrapped packets of food to take home which the child then consumes independently, one to two packets a day, at his or her own pace. However as some children may not be fed by the mother, if RUTF is the only food that a child then eats, the child may lose opportunities for psycho-social stimulation. The guidelines for community-based treatment of malnourished children identify this as an issue and promote the need to sustain psychosocial stimulation while the child is eating RUTF. Nevertheless, if there is no requirement for a young child to be fed, the degree of contact between mother and child may be reduced, perhaps with consequences for the child’s physical and cognitive development. Studies in Jamaica and Bangladesh have shown that psycho-social stimulation of malnourished children is important for their subsequent development.77-79 It would be useful to undertake long term follow-up studies of children treated at home with RUTF to assess their growth, anthropometric status, cognitive development and the risk of relapse in comparison with children treated in a conventional manner. It would be of interest to do a study comparing the rates ofrecovery of malnourished children whose mothers were givena month’s supply of either RUTF or a ration of fortified blendedfood, oil and micronutrient sprinkles so that the two rationsof food provided similar amounts of energy, fat, protein andmicronutrients. The children’s weight gain over a period of amonth could be compared and then followed by 3-6 monthsfollow up to assess the subsequent rate of growth and toassess any relapse. Tests of cognitive function could also beadministered. The hypothesis is that mothers who have toprepare food for their child and then give it may achievebetter long term growth rates because of a stronger mother-child bond and psychosocial stimulation. It would also be interesting to examine the consequencesof giving RUTF to older infants aged 6 – 12 months in termsof their long term development and any problems withmaking the transition to normal family foods.


27

Annexes9

Summary tables of published studies assessing theimpact of nutrition interventions conducted inemergency and non-emergency situations

Emergency situations1. General studies and surveys of malnutrition 2. Treatment of moderate acute malnutrition, community and facility based3. Treatment of severe acute malnutrition, community and facility based4. Micronutrients distribution5. Food security or agricultural interventions6. General ration distribution programmes 7. Conditional cash transfers or vouchers8. Cost and cost-efficiency.

Non-emergency situations1. Conditional cash transfers or vouchers2. Treatment of moderate acute malnutrition, community and facility based3. Treatment of severe acute malnutrition, community and facility based4. Micronutrients distribution

Annex 1

Annexes

Mesfin Teklu, WV, Ethiopia, 2006

AnnexesThe impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidence

28

Authorof study

Objective of Study Setting Study design,Statistics

Surveypopulation

Indicator/outcome Impact Conclusion

Seal et al(2005)

To assess the level of irondeficiency, anaemia andvit A deficiency inpopulations dependenton long-terminternational food aid &humanitarian assistance.(Very little informationhas been published ontheir prevalence on thistype of population).

Five Africanrefugeecamps inKenya,Uganda,Ethiopia andAlgeria(March2001-Sept2002).

- Observational study (cross sectional surveysin all five camps)

- Statistics presented

Children aged 6-59 monthsN= Approx 1035

Adolescentsaged 10-19y

Women aged 20-55y.

• Anaemia

• Iron deficiency

• Vitamin A deficiency

Children Prevalence of anaemiaranged from 12.8% to72.9% in different camps

Iron deficiency ranged from23% to 75%

Vitamin A deficiencyranged from 20.5% to61.7% in 4 camps assessed

Coverage Vitamin A coverage in 4camps ranged from 3.5 upto 66.2%.

The high level ofmicronutrient deficienciesseen in long-term refugeesargues in favour of furtherenhancements in food aidfortification and thestrengthening of nutritionand public healthprograms.

Seal et al(2006)

To assess the iodinestatus of long-termrefugees dependent oninternational food aidand humanitarianassistance.

Six refugeecamps inKenya,Uganda,Ethiopia,Algeria andZambia (2001-2003).

- Observational study (series of cross-sectionaltwo-stage clusteror systemic random sample surveys)


- Male and femaleadolescentsaged 10-19y

- Sample size varied from 105to 723 indifferent camps

- Urinary iodine concentration (UIC)

- Visible goitre

Median UIC ranged from254 to 1200μgl-1 and in fiveof the camps exceeded therecommended maximumlimit of 300μgl-1, indicatingexcessive iodine intake.

Visible goitre ranged 0.0 to7.1% in 4 camps.

The camp with highest UICalso had the highestprevalence of visible goitre.

The iodine conc in 11 saltsamples from 3 campsexceeded the productionlevel conc of 20-40ppmrecommended by WHO butwere < 100ppm.

Excessive consumption ofiodine is occurring in mostof the surveyedpopulations. Urgentrevision of the level of saltiodisation is required tomeet current WHOrecommendations.

However the full cause ofexcessive iodine excretionremains unknown andfurther investigation isrequired urgently toidentify the cause, assessany health impact andidentify remedial action.

Jayatissa etal (2006)

To measure the extentand severity ofmalnutrition among thetsunami affectedpopulation living in reliefcamps

Forty reliefcamps in SriLanka onemonth afterthe tsunami(Jan 17 to28, 2005)

- Observational study (A cross-sectional, two-stage, 30 cluster study). Each campwas considered a cluster and 30 children randomlyselected fromeach camp


- Children < 5y N=878

- Pregnant women N=168

- Lactating women N=97

Children- Wasting, stunting and underweight (defined as W/H, H/A and W/A below

- 2SD of the NCHS/WHO reference respectively).

Pregnant womenAcute undernutrition(defined as MUAC ≤230mm).

Lactating womenUnderweight (defined asBMI < 18.5 kg/m2) andoverweight (defined asBMI > 24.9 kg/m2).

Children16.1% wasted20.2% stunted34.7% underweight

Pregnant women37% undernourished

Lactating women31% underweight20% overweight

Although the general fooddistribution was well inplace, the food supplylacked diversity and 70.9%of the children did not getappropriate supplementaryfood.

The prevalence of bothacute and chronicundernutrition amongchildren in the camps wassignificantly higher thanthe national Sri Lankanaverage.

It was recommended thata one-day vitamin Amegadose supplementationbe given to children6months to 5y andsupplementary food begiven to children, pregnantand lactating women for aminimum of 1y in tsunamiaffected areas.

Renzaho(2007)

To assess prevalence ofmalnutrition, CMRs,causes of death, and theprevalence ofmalnutrition after 2.5yof humanitarianresponse to a crisis(chronic food insecuritycharacterised by floods in2000 and severedroughts in 2002 and2003).

(This study was conductedto provide baseline data,because soundepidemiological baselinedata were not establishedagainst which to assessthe impact of theprograms)

CahoraBassa andChangaradistricts,TeteProvince,Mozambique (2004)

Observationalstudy (a cross-sectionaltwo-stage, 30-cluster householdsurvey)

Statisticspresented

838 households

Children aged 6-59 monthsN = 874

- Acute malnutrition, W/H <-2 Z scores

- Underweight, W/A < -2 Z scores

- Chronic malnutrition, H/A < -2 Z scores

- CMRs

- Prevalence of lost pregnancies

Baseline infoNot available butanthropometric dataindicate that in 2003, theprevalence of acutemalnutrition was 9.9% inTete Province.

The prevalence of acutemalnutrition was 8.0%(95%CI: 6.2-9.8%),underweight 26.9%(95%CI: 24.0-29.9%),chronic malnutrition, 37%(95%CI: 33.8-40.2%).

CMR was 1.23/10,000/day(95%CI: 1.08-1.38) which is1.4 times higher than theCMR cut-off point used todefine excess mortality inemergencies.

Proportion of lostpregnancies was estimatedat 7.7%

The impact of the foodprogrammes would beimproved if mutuallyacceptable food aidprogramme objectives andbeneficiary targets andselection criteria aredeveloped

Emergency situations

1. General studies and surveys of malnutrition

Cont’d next page


29

Author ofstudy

Objective ofStudy

Setting Study design,Statistics

Surveypopulation

Indicator/outcome

Impact Conclusion

Johnecheckand Holland(2007)

To use data from 2large scale surveys toassess malnutritionrates and understandthe underlying causesof malnutrition in apost-conflict area

(Two large-scalestudies, the NationalSurveillance System(NSS) Pilot Study andthe National Risk andVulnerabilityAssessment (NRVA)were conducted bygovernment, UN andNGOs in Afghanistan,as part of widerefforts characterizingAfghan livelihoods inrelation to particularoutcomes of interest:vulnerability topoverty, foodinsecurity, andmalnutrition).

Afghanistan

NSS study[Nov-Dec 2003(Fall) and May-June 2004(Spring)]

NVRA study(July-Sept 2003)

NSS study -Observational study(two cross-sectionalsurveys on different HH,one in Fall and the otherin Spring)

Statistics presented

NVRA studyObservational study(one cross-sectionalsurvey)

For NSSA total of 911 and910 HH (5936 and5935 individuals ieall members in eachHH) for Fall andSpring respectively.

Children aged6 to 59 monthsN= 793 and 545 forFall and Springrespectively.

For 2003 NVRAN= 11,227 HH(85,577 individuals)

Acute malnutrition(wasting) as definedby SPHERE and theAfghan Ministry ofPublic Health is<10% GAM (W/H < -2 z-scores forchildren aged 6 to 59months).

* This aggregatedinformation is notstatisticallyrepresentative of anyregion, group, orsubgroup in theAfghan population,and should not beused for prevalenceestimates.

- Acute malnutrition rates were 8% (<80% of medianW/H) in children during thefall season and 9% duringthe spring season.

- Acute malnutrition is below emergency levels forchildren < 5y but the levelof chronic malnutritionindicates a problem ofpublic health importance.

- The findings suggest thatin addition to lack of adequate HH food intake, recurrent illness and suboptimal infant and young child feeding and hygiene practicescontribute to poor nutritional outcomes in thisage group. The survey alsofound poor access to healthcare, markets, and waterfor HH use.

Improving nutritionalstatus requires amultiprongedapproach, directlytargetingmalnutrition, coupledwith economicgrowth, HH livelihoodsecurity, socialprotection, access topublic health servicesand water andsanitation.

Olwedo et al(2008)

To estimate theprevalence of anddescribe the riskfactors for proteinenergy malnutritionamong children < 5yliving in IDP camps(as a result of aprotracted 20y civilwar)

Four IDP campsin OmoroCounty, Guludistrict NorthernUganda (2006)

- Observational study(cross-sectional surveyusing multistagesampling). Two zones ineach camp wererandomly selected. Azone constitute a studycluster


- Children aged 3to 59 monthsN= 672

Wasting andstunting using z-scores (standarddeviations) of W/Hand H/A indices.

Global refers to < -2SD.

High prevalence of globalstunting of 52.4% andglobal acute malnutritionof 6.0% were observed inthe camps.

The main sources offoodstuff for IDPs included;food rations distributed byWFP (which wereinadequate in quantity andlimited in variety),cultivation and purchase.

There is highprevalence of proteinenergy malnutrition(stunting) amongchildren in the IDPscamps in Gulu district.

Stake holders shouldintensify efforts toimprove thenutritional status ofIDPs especiallychildren in the campsettings.

Rowe et al(2008)

To obtain data onstorage, preparationand usage of fortifiedblended foodsprovided by theUSAID

(This is because animportantconsideration indetermining theability of fortifiedfood-aid commoditiesto meet the nutritionalneeds of beneficiariesis the manner in whichcommodities areutilized and preparedand the degree towhich micronutrientlosses occurs duringhandling and cookingby the beneficiaries).

Guatemala,Malawi andUganda (Jul-Aug2005)

- Observational study (interviews and physical observation)

- Some statistics presented

Primary caregiversin a household.

N= More than 100households and twowet-feeding sites

Corn-soy blend,cornmeal, soy-fortified cornmeal,soy-fortified bulgurand fortifiedvegetable oil.

Storage practises bybeneficiaries appeared tobe appropriate and allcommodities observedwere free from off-flavoursand odours.

Cooking water was typicallyobtained from boreholes oropen wells with a pH rangeof 4.7 to 7.7.

- Food preparation usually took place in covered areaswith the use of analuminium or clay pot overa wood-fuelled fire.

- Thin or thick porridges were the most common dishes prepared fromcereal-based products, withconcentration ranges of10% to 31% (wt/wt) inwater.

- Cooking times for porridges ranged from 5 to53 minutes, with a mean of26 minutes.

Cooking fuel could besaved and nutritionalquality probablyimproved if reliefagencies emphasizedshorter cooking times.These data can beused to simulatepreparation methodsin the laboratory forassessment of thenutritional impact ofcooking.

Household sizes arelarger than those usedby private voluntaryorganizations todetermine the amountof food aid allocatedto households.

Another major concernis the actual nutrientdelivery achieved afterpreparation offortified commoditiesby the beneficiaries.

Cont’d next page


30

Author ofstudy

Objective ofStudy

Setting Study design, Statistics Surveypopulation

Indicator/outcome

Impact Conclusion

Gelli et al(2009)

To estimate theprogrammaticcosts and cost-efficiencyassociated withproviding foodthrough schools infood-insecure,developing-country contexts,by analyzingglobal project datafrom the WorldFood Programme(WFP).

WFP projectdata for 2005for 42 countries

Project data, includingexpenditures and number ofschoolchildren covered, werecollected through projectreports and validated throughWFP Country Office records.Yearly project costs perschoolchild were standardizedover a set number of feedingdays and the amount of energyprovided by the average ration.Output metrics, such astonnage, calories, andmicronutrient content, wereused to assess the cost-efficiency of the dif-ferentdelivery mechanisms.

*The proportion of fooddistributed in food-for-education (FFE) activities overthe total food distribution ineach WFP project is used as aparameter that is then used toscale total project expenditure inorder to estimate the FFEexpenditure.

School agedchildren.N= > 8.6million

WFP StandardProject Report• expenditures• beneficiaries • food distribution

Estimated yearly expenditure by beneficiary

Cost-efficiency• Tonnage• Calories• Micronutrient

content delivered,

The average yearlyexpenditure per child,standardized over a 200-day on-site feeding periodand an average ration,excluding school-levelcosts, was US$21.59. Thecosts varied substantiallyaccording to choice of foodmodality, with fortifiedbiscuits provid¬ing theleast costly option of aboutUS$11 per year and take-home rations providing themost expensive option atapproximately US$52 peryear. Comparisons acrossthe different foodmodalities suggested thatfortified biscuits providethe most cost-efficientoption in terms ofmicro¬nutrient delivery(particularly vitamin A andiodine), whereas on-sitemeals appear to be moreefficient in terms of caloriesdelivered. Transportationand logistics costs were themain drivers for the highcosts.

The choice ofprogramme objectiveswill to a large degreedictate the foodmodality (biscuits,cooked meals, or take-home rations) andassociatedimplementation costs.Fortified biscuits canprovide substantialnutritional inputs at afraction of the cost ofschool meals, makingthem an appealingoption for servicedelivery in food-insecure contexts.Both costs and effectsshould be consideredcarefully whendesign¬ing theappropriate school-based intervention.The costs estimates inthis analysis do notinclude all school-levelcosts and are thereforelower-boundestimates of fullimplementation costs.

Hossain et al(2009)

To assess therelationshipbetween food aidand acutemalnutritionfollowing the Oct2005 earthquakein Pakistan

Mansehra andMuzaffarabaddistricts,northernPakistan.(Mansehra, 21to 25 Nov;Muzaffarabad,14 to 18 Dec2005)

- Observational study (two separate cross-sectionalsurveys, one in Mansehra district and the other in Muzaffarabad district). Clusterswere selected using theprobability proportional to size(PPS) methodology.


- Children aged6 to 59.9months

- Mansehra N= 602

- Muzaffarabad N= 512

- Acute malnutritiondefined as wasting (WHZ) < -2 SD basedon WHO/NCHS/CDCP reference orpresence of oedema.

Informationcollected included;• Receipt of food aid• Child health• Level of HH

damage• Water sources and• Excreta disposal

Children in HH receivingfood aid had a lowerprevalence of acutemalnutrition in Mansehra(adjusted OR, controlled forconfounders, 0.40; 95%CI,0.13 to 1.22) andMuzaffarabad (adjustedOR, 0.72; 95% CI, 0.36 to1.44).

The coverage of food aidtended to be higher inclusters with a greaterpercentage of homescompletely destroyed.

Food aid recipientshad lower prevalenceof acute malnutritionthan HH not receivingfood aid.

Communities withhigher levels ofcomplete HHdestruction were morelikely to receive foodaid suggesting thatfood aid wassuccessfully targetedto the areas ingreatest need.

Shrimpton etal (2009)

To assess theadequacy ofimproved generalfood ration inreducing theincidence of lowbirthweight inBhutanese refugeecamps in Nepal.

This research wascommissioned byWFP becausereports indicatedthat rates of lowbirthweight in thecamps were closeto 10% ascompared with15% for Bhutan asa whole, 21% forNepal and >30%in hospitalssurrounding campareas.

Four Bhutaneserefugee campsin Nepal (2009)

- Retrospective study of medical records on birthweightand maternal factors from 1994 to 2001 were analysed.

- The adequacy of the food ration and food supplementsprovided to pregnant womenwere calculated using food-composition tables for 4 different periods; 1993-94,1995-97, 1998, 1999-2001 andassessed by comparing the nutrient energy density of theration provided to the generalpopulation both with andwithout the supplementprovided to women during pregnancy in the four periods


InfantsN= 5,672

MothersN=5,234

- Low birthweight < 2.5kg

- Nutritional adequacy of food ration provided to all refugees and the supplements given to pregnant mothers were also evaluated.

The rates of lowbirthweight were low inthe refugee camps,averaging 11% in the yearsreviewed.Between 1996 and 1998,the mean rates of lowbirthweight fell from 16%to 8% and meanbirthweight increased from2.84kg (SE, 2.80-2.87) to3.0kg (SE, 2.97-3.03).

- The increase inbirthweight occurredfollowing improvements inthe micronutrient-to-energy ratios of the generalration (from mid 1994)

Rates of lowbirthweight wereprobably achievedbecause basic needs ofmothers were met,including both thequantity and themicronutrient contentof food, water andsanitation, antenatalcare and education.

It is hypothesized thatincreasedpericonceptionalmicronutrient intakemay be responsible forthe increase inbirthweight.

Cont’d next page


31

Author ofstudy

Objective ofStudy


Surveypopulation

Indicator/outcome

Impact Conclusion

Chotard et al(2010)

To estimate levels of andfluctuations in wastingprevalence in childrenfrom surveys conductedin arid and semiaridareas of the GreaterHorn of Africa accordingto livelihood.

(The prevalence ofwasting in preschoolchildren is widely used toassess possible needs foremergency humanitarianinterventions in areasvulnerable to drought,displacement and relatedcauses of food insecurity).

Eritrea, Ethiopia,Kenya, Somalia,Southern Sudanand Uganda(2000-2006)

- Observational study (results from around 900 area-level nutrition surveys were compiled and analyzed)


Children aged 0to 5 y

LivelihoodpopulationsPastoral (PP)Agricultural (AP)Mixed (MP)Migrants (MiP)

Wasting defined as < -2 SD weight-for-height).

Fluctuations in wasting weregreater among pastoralpopulation during years ofdrought, with prevalencerising to 25% or higher;prevalence amongagricultural populationseldom exceeded 15%.

Among pastoral population,the average prevalence ofwasting was about 17%, 6-7percentage points higherthan the rates amongagricultural population ormixed population.

IDPs and urban migrants havesomewhat higher prevalencerates of wasting.

Tracking changes inwasting prevalenceover time at the arealevel–eg with time-series graphicalpresentations–facilitatesinterpretation ofsurvey resultsobtained at any giventime.Different populationshould be judged bypopulation-specificcriteria and invariantprevalence cut-offpoints avoided;interpretation rulesare suggested.


32

Author ofstudy

Objective ofStudy


Surveypopulation

Indicator /outcome

Impact Conclusion

Nielson et al(2004)

To investigate theeffect of asupplementaryfeedingprogramme (SFP)on malnourishedchildren inGuinea-Bissauwho werereturning to theirhomes afterhaving beendisplaced withinthe countrybecause of war in1998–1999.

Bissau, GuineaBissau (Sep 1,1998- May 31,1999)

Observational study.The effect of the war onthe nutritional status ofchildren who werepresent in Bissau wasevaluated by comparingthe mortality and theprevalence ofmalnutrition with thevalues expected had thewar not occurred and bycomparing the severity ofmalnutrition inmalnourished childrenbefore and during thewar.

- No control group


Children < 5ywith MUAC <130mm.N= 433 of which247 receivedsupplementaryfeeding.Children withMUAC < 130 mmwere providedweekly medicalconsultationsandsupplementaryfeeding for 7d oruntil recovery(MUAC ≥130mm)

Mortality

Recovery rate

Growth rate usingMUAC

Weight gain

Baseline infoMedian (quartiles) ofMUAC (mm) = 122(118, 126).

The degree of malnutrition andmortality of malnourished childrendid not increase during the war. Theprevalence of malnutritionincreased with the start of the warbut then decreased. Seventy-fourpercent of the referred childrenreceived treatment (N=247); ofthose, 1% died, 67% (95% CI: 61%,73%) recovered, and 32% (95% CI:26%, 38%) abandoned treatment.Compliance was 89%. The recoveryrate was 13.1/1000/d, and themedian time to recovery was 48 d(95% CI: 34, 72 d). MUAC growthrate of 0.78 (95%CI: 0.31, 1.25)mm/d and weight gain of 4 (95%CI:3, 5) g/d was observed. Bettercompliance was associated withshorter time to recovery.

Our findings may bebiased by changes inthe cultural andsocioeconomicbackground of themalnourishedchildren. However, 3different analysesindicated abeneficial effect ofthe SFP.Thus, the home-based SFP probablypreventednutritionaldeterioration duringthe war in Guinea-Bissau.

Maleta et al(2004)

To compare theeffect of maizeand soy flourwith that ofready-to-usefood (RTUF) inthe hometreatment ofmoderatelymalnourishedchildren in anarea with chronicfood insecurityand poverty.

Lungwena,Mangochidistrict, Malawi(Jul 2000 - April2001)

A parallel groupintervention study.

Intervention - one of twofood supplements: RTUF(92 g/day) or maize andsoy flour (140 g/day) or 12weeks. Both supplementsprovided 2 MJ (500Kcal) ofenergy daily but haddifferent energy andnutrient densities.

There was a 12-week postsupplementationobservation. Weight andheight were measured at4-week intervals duringintervention and at theend of the 12-weekfollow-up. Twenty-four–hour dietaryrecall was assessedfortnightly

• No control group

• Statistics presented

Underweightand stuntedchildren aged 42to 60 mo (WAZand HAZ <-2,WHZ < -0.3)

N= 61 of whichRTUF= 30 andmaize and soyflour= 31

- Participantswere drawnfrom theLungwena ChildSurvival Study(LCSS) cohort.

- Weight and height gain

- Dietary intake

Baseline infoMean (SD) of WHZ,WAZ and HAZ was −0.9 (0.5), −2.5(0.5) and −3.2 (0.9)in the RUTF groupand −0.9 (0.5), −2.4(0.5) and −3.0 (0.8)in the maize and soyflour grouprespectively

During the supplementation phase,the consumption of staple food fellamong children receiving maize andsoy flour but not among thosereceiving RTUF. There was thushigher intake of energy, fat, iron,and zinc in the RTUF group.

Both supplements resulted inmodest weight gain, but the effectlasted longer after RTUFsupplementation [mean diff = 77(95% CI: 20, 140) p= 0.01].

Height gain was not affected ineither group. Periodic 24-hourdietary recalls suggested that thechildren received only 30% and43%, respectively, of thesupplementary RTUF and maize andsoy flour provided.

RTUF is anacceptablealternative to maizeand soy flour fordietarysupplementation ofmoderatelymalnourishedchildren.Approaches aimedat increasing theconsumption ofsupplementary foodby the selectedrecipients areneeded.

Patel et al(2005)

To test thehypothesis thatsupplementalfeeding withRUTF wouldresult in bettergrowth inMalawianchildren at risk ofmalnutritionthan feedingwith corn/soy-blend(Crises in Malawiaffects foodproduction hencechronic foodinsecurity exists)

Sevennutritionalrehabilitationunits (NRUs) inrural areas ofsouthernMalawi (Dec 2002 toMay 2003)

A controlled, comparativeclinical effectiveness trialof 2 supplementaryregimens. A stepped-wedge design withprospective systematicallocation was used forassigning children toreceive either ready-to-use therapeutic food(RUTF) or micronutrient-fortified corn/soy-blend(MFCSB) for up to eightweeks.

- No control group


Children aged10-60 mo at riskof malnutrition,W/H >80%, but< 85% of thestandardreference valuesof WHO withoutoedema.

N= 331 (RUTFgroup)N= 41 (MFCSBgroup)

- Recovery (defined as W/H > 90%)

- Rate of weight gain.

Baseline infoMean ± SD of W/H(% of internationalstandard)= 82 ± 1,and 82 ± 2 for MFCSBand RUTF groupsrespectively

Children receiving RUTF were morelikely to recover (58% vs 22%;difference 36%; 95% CI: 20- 52) andhad greater rates of weight gain(3.1 g/kg.d vs 1.4 g/kg.d; difference1.7; 95% CI 0.8-2.6) than childrenreceiving corn/soy-blend.

*Main limitation of this study wasthat its participants were notrandomly assigned to receive eithercorn/soy-blend or RUTF.

The results of thispreliminary worksuggest thatsupplementaryfeeding with RUTFpromotes bettergrowth in children atrisk of malnutritionthan the standardfortifiedcereal/legume-blended food.

2. Summary of published studies assessing the impact of moderate acute malnutrition programmes in emergency situations

Cont’d next page


33

Authorof study

Objective ofStudy


Indicator /outcome

Impact Conclusion

Ruel et al(2008)

To compare the effectof (i) targeting allchildren aged 6-23mo (preventive) andii) targetingunderweightchildren youngerthan 5 years(recuperative) inreducing communityprevalence ofstunting,underweight, andwasting in a food-assisted maternaland child health andnutrition programmein Haiti.

(Previous evidencesuggests thattargeting nutritioninterventions earlierin life, before childrenbecomeundernourished,might be moreeffective for reductionof childhoodundernutrition).

Threecommunes ofCentralPlateauregion in Haiti(May-Sep,2002 andexactly 3years later)

- A cluster randomised trial. - Compared two programmes which included a behaviour change and communication (BCC) component: a preventivemodel and a recuperativemodel. Both models alsotargeted pregnant andlactating women.

Two cross-sectional surveys (atbaseline and 3 years later)

Intervention = 8kg/mo wheat-soy blend + oil general ration,health education, growthmonitoring, parasitestreatment, immunizations, vitA supplement, ORS provisionand home visits

- Duration of intervention was 9 mo in recuperative and up to18 mo for preventive group.

- The two programme models were delivered at the community level, rather than the individual level.

- No control group


Preventive groupAll children aged 6-23 mo.

N= 10 communities

Recuperative group Underweightchildren (WAZ < -2)aged 6-59 mo andchildren aged 24-59mo with WAZ<-3.

N= 10 communities

Children aged 12–41 months wereselected for effectassessment(roughly 1500children persurvey).

• Mean Z scores for H/A, W/A and W/H

• Prevalence of childhood stunting, underweight, and wasting (using WHO 2006 reference

data).

Baseline infoThere were nodifferencesbetweenprogrammegroups atbaseline.

At follow-up, stunting,underweight, and wasting were4–6 percentage points lower inpreventive than in recuperativecommunities; and meananthropometric indicators werehigher by +0·14 z scores (heightfor age; p=0·07), and +0·24 zscores (weight for age and weightfor height; p<0·0001). The effectwas greater in children exposed tothe preventive programme for thefull span between 6 and 23months of age than in childrenexposed for shorter durationsduring this period. The quality ofimplementation did not differbetween the two programmes;nor did use of services formaternal and child health andnutrition.

The preventiveprogrammewas moreeffective for thereduction ofchildhoodundernutritionthan thetraditionalrecuperativemodel.

Matilsky etal (2009)

To determine therelative recovery andgrowth rates formoderately wastedchildren receivingeither soy/peanut FS,milk/peanut FS, orCSB as asupplementary food.

The sites wereidentified in thesouthern region ofMalawi based oncensus reports ofmoderately wastedchildren provided bythe World FoodProgramme.

Twelve ruralstudy sites,southernMalawi (Jul 2007 toFeb 2008)

A randomized clinicaleffectiveness trial of 3 locallyproduced foods for 8 weeks(reassessment and supply ofsupplement food wasperformed every two weeks).

- Children were randomly assigned to 1 of the 3 diets: milk/peanut FS, soy/peanut FS,or CSB

- Focus group discussions wereused to collect information from caretakers.

- No control group


Children aged 6–60mo with moderatewasting (defined asWHZ < -2 but ≥ -3without oedema)

N= 1362 of whichmilk/peanut FS n=465, soy/peanut FSn= 450 and CSB n=447

- Recovery, defined as havinga WHZ > -2

- Weight gain rate

- Adverse outcomes such asseveremalnutrition ordeath.

- Cost

Baseline infoThe mean (SD) ofWHZ was -2.2 ±0.4 for all threegroups.

Mean weight (kg)was 7.56 ± 1.68,7.55 ± 1.77, 7.46± 1.80 formilk/peanut FS,soy/peanut FSand CSBrespectively.

- Children receiving soy/peanutFS had a similar recovery rate tothose receiving milk/peanut FS andchildren in either FS group weremore likely to recover than thosereceiving CSB (80% in both FSgroups vs. 72% in the CSB group;P< 0.01).

- The rate of weight gain in thefirst 2 wk was greater amongchildren receiving milk/peanut FS(2.6 g/kg/d) or children receivingsoy/peanut FS (2.4 g/kg/d) thanamong children receiving CSB (2.0g/kg/d; P < 0.05).

- Rates of length gain did notdiffer among the 3 groups.

- A total of 8% of children in each feeding group developed malnutrition, while receiving supplemental feeding.

- The cost of the foods in Malawiat the time of the study for locally manufactured milk/peanut FS wasUS$ 0.16/1000 kJ, soy/peanut FSwas US$ 0.08/1000 kJ, and CSBwas US$ 0.04/1000 kJ.

We concludethat FS aresuperiorsupplementaryfoods to CSB formoderatelywastedMalawianchildren.

Cont’d next page


34

Authorof study

Objective ofStudy


Surveypopulation

Indicator / outcome Impact Conclusion

Pérez-Expósitoand Klein(2009)

A review toexamine theimpact thatfortified blendedfoods used inhumanitarianrelief programshave had on thehealth andnutritional statusof infants andyoung childrenwith moderatemalnutrition, or atrisk ofundernutrition, indevelopingcountries.

Variousdevelopingcountries(studyperiod notstated)

Published articles wereidentified usingelectronic databases ofPubMed and theCochrane Library withno restrictionsregarding date, age, orlanguage as well asgeneral Web searches.Search terms includedcommodity types andnames and termsrelated to foodassistance andfortification programs.

Published studiesand reports assessingeither the efficacy oreffectiveness of FBFsto improve thehealth andnutritional status ofinfants and youngchildren withmoderatemalnutrition, or atrisk of undernutritionin developingcountries

N= 11 publishedstudies and reports

- Recovery from moderate acute malnutrition, defined as weight-for-height z score (WHZ)> -2 or weight-for-height > 85%of the median + no oedema hasbeen used as an indicator ofprogramme success in many ofthe supplementary feeding interventions.

- Weight gain

Positive effects on recoveryfrom moderate acutemalnutrition and weightgain were observed whenfortified blended foodswere distributed as dietarysupplements.

Prevention of severemicronutrient deficienciesin populations reliant onfood aid has been reported,but measurements ofmicronutrient status haverarely been conducted.

Evidence of theefficacy of fortifiedblended foods forimprovingnutritionaloutcomes iscurrently limitedand weak.

Defourny etal (2009)

To evaluate alarge-scaledistribution(blanketsupplementaryfeeding) of anutritionalsupplement on theprevention ofwasting in childrenliving in areas withhigh burden ofchild malnutritionand chronic foodinsecurity(especially duringthe hunger gapperiod)

Maradi,Niger (Mayto Oct2007)

- Intervention study

Intervention consistedof a monthlydistribution of 4 pots ofready-to-use food(RUF) (325 g/pot) toeach child for sixmonths.At each monthlydistribution, childrenwere also screened forSAM and referred fortherapeutic treatmentif indicated.

- (No clear control group but in one district children admitted to the therapeuticprogramme andregistered for RUFdistribution werecompared with childrennot registered for distribution but had been admitted for therapeutic programme)


All children aged 6 to36 mo (or heightbetween 60-85cm)with moderatemalnutrition (MUAC> 110mm)N= approximately60,000

Children with MUAC< 110mm wasreferred fortherapeutic feeding

MUAC < 110 mm

Baseline infoData for 2002-2005 presented asa figure

Cont’d impact for this studyThe proportion of complicatedcases of malnutrition requiringintensive care was lower amongchildren registered for RUFdistribution (7.8%, p= 0.005)than among those notregistered for distribution(9.9%). The proportion ofmalnourished children treatedexclusively in outpatientfacilities was higher amongbeneficiaries of the preventivedistribution (84.2%, p<0.0001)than among children notregistered for distribution(80.5%).Among children receiving apreventive distribution of RUF,cure rate was higher (92.3%, p=0.003) and default rate waslower (4.7%, p = 0.026), whencompared with rates of childrennot receiving any preventivedistribution (cure rate: 90.1%;and default rate: 6%).

Admission trends for severewasting (WFH < 70% NCHS)in Maradi, 2002–2005 showan increase every yearduring the hunger gap. Incontrast, in 2007,throughout the period ofthe distribution, theincidence of severe acutemalnutrition (MUAC < 110mm) remained at extremelylow levels. Comparison ofyear-over-year admissionsto the therapeutic feedingprogramme shows that the2007 blanket distributionhad essentially the sameflattening effect on theseasonal rise in admissionsas the 2006 individualizedtreatment of almost 60,000children moderatelywasted.

These resultsdemonstrate thepotential fordistribution offortified spreads toreduce theincidence of severewasting in largepopulation ofchildren 6–36months of age.Although furtherinformation isneeded on thecost-effectivenessof suchdistributions, theseresults highlightthe importance ofre-evaluatingcurrent nutritionalstrategies andinternationalrecommendationsfor high burdenareas of childhoodmalnutrition.

Cont’d next page


35

Author ofstudy

Objectiveof Study


Surveypopulation


Isanaka et al(2009)

To evaluate theeffectiveness ofa population-based,preventivedistribution ofready-to-usetherapeuticfood (RUTF) onthe nutritionalstatus,mortality andmorbidity ofchildren inNiger (an areawith high levelsof childmalnutritionand chronicfood insecurity)

Twelvevillages inMaradi,Niger(August2006 -March2007).

- A cluster randomized trial of12 villages, 6villages were randomized to intervention and 6to no intervention.

Interventionvillages received amonthlydistribution of 1packet per day ofRUTF (92g [500kcal/d]) for3months (Aug toOct 2006).

- Non intervention villages received nopreventive supplementation.

Follow-up periodwas over 8 month.


Non malnourishedchildren with weight-for-height 80% or moreof the NCHS referencemedian aged 6 to 60months

N= 1671 forintervention group

N= 1862 for nonintervention group

- Changes in WHZ accordingto the WHO Child Growth Standards

- Incidence of wasting (WHZ < −2) over the 8 months of follow-up.

Baseline infoIntervention groupThe mean (SD) for WHZ was-0.7 (1.0) and HAZ was -1.9

(1.3). 8.2% and 1.0% werewasted and severely wastedwhile 46.7% and 17.6%were stunted and severelystunted respectively.

Non interventionThe mean (SD) for WHZ was -0.7 (1.0) and HAZ was -2.2(1.3). 8.6% and 1.4% werewasted and severely wastedwhile 56.2% and 27.9%were stunted and severelystunted respectively.

The number of children with heightand weight measurements in August,October, December and February was3166, 3110, 2936, and 3026,respectively. The WHZ differencebetween the intervention and nonintervention groups was −0.10 z(95% CI: −0.23 to 0.03) at baselineand 0.12 z (95% CI: 0.02 to 0.21) after8 months of follow-up. The adjustedeffect of the intervention on WHZfrom baseline to the end of follow-upwas thus 0.22 z (95% CI, 0.13 to 0.30).The absolute rate of wasting andsevere wasting, respectively, was 0.17events per child-year (140 events/841child years) and 0.03 events per child-year (29 events/943 child-years) in theintervention villages, compared with0.26 events per child-year (233events/895 child-years) and 0.07events per child-year (71 events/1029child-years) in the non interventionvillages. The intervention thusresulted in a 36% (95% CI, 17% to50%; P< 0.001) reduction in theincidence of wasting and a 58% (95%CI, 43% to 68%; P< 0.001) reductionin the incidence of severe wasting.There was no reduction in mortality,with a mortality rate of 0.007 deathsper child-year (7 deaths/986 child-years) in the intervention villages and0.016 deaths per child-year (18deaths/1099 child-years) in the nonintervention villages (adjusted hazardratio, 0.51; 95% CI, 0.25 to 1.05).

Short-termsupplementation of nonmalnourishedchildren withRUTF reducedthe decline inWHZ and theincidence ofwasting andsevere wastingover 8 months.

Isanaka et al(2010)

To compare theincidence ofwasting,stunting, andmortalityamong childrenwho arereceivingpreventivesupplementation with eitherready-to-usesupplementaryfoods (RUSFs)or ready-to-usetherapeuticfoods (RUTFs)in an area withhigh levels ofchildmalnutritionand chronicfood insecurity

Twelvevillages inMaradi,Niger (Apr2007 – Oct2007).

Population-basedstudy- Cohort study (two groups)

- No control group.

- Children either received a monthly distribution ofRUSFs (247 kcal [3spoons] per day) for6 months or RUTFs(500-kcal sachet perday) for 4 months.

- Distribution of supplements and follow ups were done monthly.


All children aged 6 to 36months in the 12villagesN= 1645 of which 873received RUTFs and 772received RUSFs.

*Study limitations In addition to the dose,the 2 preventivestrategies undercomparison differed inimportant ways,including the duration ofsupplementation, modeand time of initiation ofdistributions, as well asthe age of childreneligible forsupplementation.

The frequency ofanthropometricscreening also differedaccording to strategy;children who received theRUSF strategy werescreened twice as oftenas those who received theRUTF strategy, because ofscreening at both theRUSF distribution sitesand monthly follow-upvisits. As a result, ourconclusions relate to therelative performance ofthe 2 preventivestrategies overall ratherthan to the individualproducts.

Incidence of wasting (WHZ<-2), severe wasting (WHZ < -3), stunting (HAZ < -2), andsevere stunting (HAZ < -3)according to WHO growthstandards, and mortality.

Baseline infoData shown as mean (SD)

RUTF groupWHZ= -0.7 (1.1). 11.8%were wasted of which 3.2%were severely wasted.HAZ= -2.4 (1.1). 64.6% werestunted of which 31.9%were severely stunted.

RUSF groupWHZ= -1.0 (1.1). 16.4%were wasted of which 3.7%were severely wasted.HAZ= -2.4 (1.1). 65.8% werestunted of which 26.8%were severely stunted.

More children in the RUSFstrategy than in the RUTFstrategy were in villagesthat received the previousnutritional intervention(50.3% vs 43.9%). Onaverage, children had higherWHZs (-0.76 ±1.07 vs -0.93±1.10) and HAZs (-2.24±1.04 vs -2.56 ±1.15) atbaseline in villages in whichnutritional intervention waspreviously implemented.

The effectiveness of RUSFsupplementation depended onreceipt of a previous preventiveintervention. In villages in which apreventive supplementationprogramme was previouslyimplemented, the RUSF strategy wasassociated with a 46% (95%confidence interval [CI]: 6%–69%)and 59% (95% CI: 17%–80%)reduction in wasting and severewasting, respectively. In contrast, invillages in which the previousintervention was not implemented,we found no difference in theincidence of wasting or severewasting according to type ofsupplementation.Compared with the RUTF strategy, theRUSF strategy was associated with a19% (95% CI: 0%–34%) reduction instunting overall.

We found thatthe relativeperformance ofa 6-month RUSFsupplementation strategyversus a 4-month RUTFstrategy variedwith receipt of apreviousnutritionalintervention.Contextualfactors willcontinue to beimportant indetermining thedose andduration ofsupplementation that will bemost effective,acceptable, andsustainable fora given setting.

Cont’d next page


36

Author ofstudy

Objective ofStudy


Surveypopulation


Nackers et al(2010)

To compare theclinicaleffectiveness ofready-to-usetherapeutic food(RUTF) and acorn/soy-blend(CSB) based pre-mix for thetreatment ofmoderatelymalnourishedchildren in Niger.An area affectedby chronic foodinsecuritymarked each yearwith a ‘hungergap’ period whenprevious year’sstocks have runout.

Two SFCs inZinderregion,South ofNiger(Aug 2007to July2008)

Field-randomized trial.Children were individuallyrandomized to;

Group 1Receive RUTF (Plumpy’Nut)two packs daily, i.e. 1000kcal/day]

Group 2Receive a CSB pre-mix(1750g of CSB, 175 ml ofvegetable oil and 105 g ofsugar, i.e. 1231 kcal/day).

• Children were followed weekly up to recovery (recruitment took place from Aug-Oct 2007 with a 6-mo follow up).

• No control group


All childrenmeasuring 65 to<110 cm (used as aproxy for the age of6–59 mo), newlyadmitted to the SFCswith MAM and goodappetite. MAM was defined as aW/H% of the medianfrom 70 to <80%(NCHS ref), withoutoedema and withMUAC ≥110 mm.

N= 215 in RUTF groupN= 236 in CSB pre-mix group.

- Weight gain (g/kg/day)

- Recovery rate

- Also transfer to the inpatient TFC (I-TFC), mortality, non-responderand defaulter rates, lengthof stay, MUAC gain andhaemoglobin gain duringtreatment, relapse andheight gain 6 months afterdischarge.

Baseline infoData are mean±SD

RUTF groupW/H %median= 76.7 ± 2.6MUAC (mm)= 120.9 ± 4.9

CSB pre-mix groupW/H %median= 76.4 ± 2.4MUAC (mm)= 120.0 ± 5.6

Overall recovery rate was79.1% and 64.4% in the RUTFand CSB pre-mix grouprespectively (p<0.001).There was no evidence for adifference between death,defaulter and non-responderrates. More transfers to the I-TFC were observed in the CSBpre-mix group (19.1%)compared to the RUTF group(9.3%) (p= 0.003). The average weight gain up todischarge was 1.08 g /kg/dayhigher in the RUTF group (95%CI: 0.46–1.70) and the lengthof stay was 2 weeks shorter inthe RUTF group (p<0.001).

For thetreatment ofchildhood MAMin Niger, RUTFresulted in ahigher weightgain, a higherrecovery rate, ashorter lengthof stay and alower transferrate to the I-TFCcompared to aCSB pre-mix.This might haveimportantimplications onthe efficacy andthe quality ofSFPs.

Thakwalakwa et al (2010)

To determinewhether lipid-based nutrientsupplements(LNS) or corn-soyblend (CSB)supplementationimproves weightgain ofmoderatelyunderweightchildren in anarea with highlevels of childmalnutrition andchronic foodinsecurity.

Lungwena,Mangochidistrict ofMalawi (Dec 2006to Feb2007).Follow upof lastparticipantended inMay 2007.

- A single-centre randomised, controlled, investigator-blinded clinical trial for 12 weeks.

- Trial interventionControl group: no food supplement CSB group: received 500g of CSB weekly LNS group: received300g of LNS weekly for 12weeks. Food supplementswere delivered to theirhomes. Both supplementswere fortified with micronutrients although levels of fortification varied between products.


Moderatelyunderweight infants& children aged 6 to15 mo with WAZ < -2based on NCHS/CDCgrowth reference.

N= 192 of which 59-control group; 67- CSBgroup & 66- LNSgroup.

Weight change

Changes in anthropometricindices, Hb levels andmorbidity.

Baseline infoValues are presented asmean (SD).Weight (kg) and WAZ forcontrol, CSB & LNS were 7.03± 0.74 and -2.98±0.86; 6.89±1.04 and -3.06 ±1.08; &7.11 ±0.91 and -2.83 ±0.78respectively.

Body weight increases did notdiffer and were 620 ±470, 510±350, and 470 ± 350 g in theLNS, CBS and control groupsrespectively (p= 0.11).

Compared with controls,infants and children in the LNSgroup gained more weight[mean (95%CI) = 150g (0-300g); p= 0.05] and had agreater increase in WAZ [0.33 (-0.02-0.65); p= 0.04].Weight and WAZ changes didnot differ between the controland CSB groups.In exploratory stratifiedanalysis, the weight increasewas higher in the LNS groupcompared with the controlgroup among those with lowerinitial WAZ [250g (60- 430g;p= 0.01].

Supplementation with LNS butnot CSBmodestlyincreasesweight gainamongmoderatelyunderweightchildren and theeffect appearsmostpronouncedamong thosewith a lowerinitial WAZ.


37

3. Summary of published studies assessing the impact of severe acute malnutrition programmes in emergency situations

Author ofstudy

Objectiveof Study


Indicator /outcome

Impact Conclusion

Sandige et al(2004)

To determinethe efficacy ofhome-basedtherapy withready-to-usefood (RTUF) inproducingcatch-upgrowth inmalnourishedchildren and tocompare locallyproduced RTUFwith importedRTUF for thispurpose.

Blantyre, Malawi (Apr – Aug 2002)

(The area has highrates of childmalnutrition (ofwhich asignificantfraction are HIVinfected) andchronic foodinsecurity.Children with HIVinfection inMalawi do notreceiveantiretroviraltherapy due toeconomicconstraints uponthe healthcareservices).

- Observational intervention study

Children were systematicallyallocated to receive hometherapy with either importedcommercially produced RTUF orlocally produced RTUF after abrief inpatient stabilization at ahospital.

- Each child received73kJ/kg/day and was followedup fortnightly. Analyses were stratified by human immunodeficiency virus (HIV) status.

Intervention lasted for 16 weeks

- Reassessment was performed every two weeks.

- No control group


Severelymalnourished(WHZ < -2 withor withoutoedema)children aged 1to 5y. N= 260 of whichlocal RTUF= 135and importedRTUF= 125

Recovery(attaining WHZ >−0.5 by 14 weeks)

- Relapse, death,or failure toachieve WHZ >−0.5 after 16weeks.

Baseline infoMean ± SD of WHZwas −2.1 ± 0.8and −2.2 ± 0.9 inlocal and importedgroupsrespectively.

Seventy eight percent of all childrenreached WHZ > −0.5, 95% of thosewith HIV-negative status and 59%of those with HIV-positive status.Eighty percent of those receivinglocally produced RTUF and 75% ofthose receiving imported RTUFreached WHZ > −0.5. Thedifference between recovery rateswas 5% (95% CI: −5–15%). The rateof weight gain was 0.4 g/kg/day(95% CI, −0.6, 1.4) greater amongchildren receiving locally producedRTUF. The prevalence of diarrhoeareported by mothers was 3.7% forlocally produced RTUF and 4.3% forimported RTUF. After completion ofhome therapy and resumption ofhabitual diet for 6 months, 91% ofall children maintained a normalWHZ.

Home-basedtherapy withRTUF wassuccessful inaffectingcomplete catch-up growth. Inthis study,locally producedand importedRTUF weresimilar inefficacy intreating ofseverechildhoodmalnutrition

Manary et al(2004)

To test thehypothesis thatthe recoveryrate for ready-to-use food(RTUF) isgreater thantwo otherhome baseddietaryregimens in thetreatment ofmalnutrition.

Blantyre, Malawi (Jan 25 – Oct 15,2001)

(The area has highrates of childmalnutrition andchronic foodinsecurity. Thestandardtreatment ofseveremalnutrition inMalawi oftenutilises prolongedinpatient care,and afterdischarge resultsin high rates ofrelapse).

- Observational interventionstudy

- Children were systematicallyallocated to one of three dietaryregimens: RTUF (730 kJ/kg/day),RTUF supplement (2100 kJ/day),or blended maize/soy flour (730kJ/kg/day).

- Children were followedfortnightly.

- Children completed the studywhen they reached WHZ >0,relapsed, or died.

- No control group


HIV negativechildren >1ydischarged fromthe NRU.N= 282 of whichRTUF group= 69,RTUFsupplementgroup= 96 andmaize/soygroup= 117

- Recovery (WHZ> 0, WHZ, basedon initial height).- Weight gain

Baseline infoNo significantdifferencesbetween groups

Mean (SD) WHZwas -1.8 (0.8), -2.0(0.9) and -1.9 (1.0)for RUTF, RUTFsupplement andmaize/soy groupsrespectively.

- Children receiving RTUF weremore likely to reach WHZ >0 thanthose receiving RTUF supplement ormaize/soy flour (95% v 78%, RR 1.2,95% CI: 1.1 to 1.3).

- The average weight gain was 5.2g/kg/day in the RTUF groupcompared to 3.1 g/kg/day for themaize/soy and RTUF supplementgroups. Six months later, 96% of all childrenthat reached WHZ > 0 were notwasted.

*The primary limitation of thestudy was that dietary intake wasnot measured at home, and thus weare uncertain as to why RTUF was asuperior home therapy.

Home basedtherapy ofmalnutritionwith RTUF wassuccessful;furtheroperationalwork is neededto implementthis promisingtherapy.

Ndekha et al(2005)

To determine ifhome-basednutritionaltherapy willbenefit asignificantfraction ofmalnourished,HIV-infectedMalawianchildren, and todetermine ifready-to-usetherapeuticfood (RUTF) ismore effectivein home-basednutritionaltherapy thantraditionalfoods.

Blantyre, Malawi (Jan – Sep 2001)

(The area hashigh rates of childmalnutrition andchronic foodinsecurity.Children with HIVinfection inMalawi do notreceiveantiretroviraltherapy due toeconomicconstraints uponthe healthcareservices).


- Children were prospectively,systematically allocated toreceive RTUF (730 kJ/kg/day),RTUF supplement (2100 kJ/day),or blended maize/soy flour (730kJ/kg/day) based on their weekof discharge from the hospital

- Children did not receiveantiretroviral chemotherapy

- Children were followedfortnightly. Children completedthe study when they reached100% weight-for-height,relapsed or died.

- Follow up was after 6 months

- No control group


HIV-positiveseverelymalnourishedchildren aged12-60 moN= 93 of whichRTUF group= 20,RTUFsupplementgroup= 28 andmaize/soygroup= 45

- Reaching 100%weight-for-height

- Weight gain rate

Baseline infoMean (SD) WHZwas -2.0 (1.1), -2.8(0.9) and -1.8 (0.8)for RUTF, RUTFsupplement andmaize/soy groupsrespectively.

Fifty-two children (56%) of allchildren reached 100% weight-for-height.Eleven children died

Median survival from the time ofenrolment in the home-basedtherapy study for these childrenwas 42 d.

Regression modeling found that thechildren receiving RUTF gainedweight more rapidly and were morelikely to reach 100% weight-for-height than the other two dietarygroups (p< 0.05).

*An important limitation of thisstudy was the differences in severityof malnutrition between thedifferent dietary groups. The childrenreceiving RUTF supplement wereclearly more wasted

More than halfofmalnourished,HIV-infectedchildren notreceivingantiretroviralchemotherapybenefit fromhome-basednutritionalrehabilitation.Home-basedtherapy RUTF isassociated withmore rapidweight gain anda higherlikelihood ofreaching 100%weight-for-height.

Cont’d next page


38

Author ofstudy

Objective ofStudy


Survey population Indicator /outcome

Impact Conclusion

Kumar andBhawani(2005)

To assess the impactof ‘Nutrition CareCentres’ onnutritional status in achronic moderate tosevere droughtaffected area.

Villages in thedistrict ofRajasthan, India(March and Sept,2003)

- Intervention study.Intervention =provision of targetedfeeding and care atnutrition care centres(NCC).Impact was measuredafter 6mo ofintervention

- No control group

- Some statisticspresented

- All severely andmoderately malnourishedchildren, W/A or W/H < 2SD below NCHS medianreference, aged <5y.

Initial survey of all childrenin 40 selected villagesN=3,206

- Follow up survey in 22intervention villages after6 monthsN= 1775

Severe underweight(defined as W/A < 3SD below NCHSmedian reference).

Baseline infoAbout 66.7% ofchildren wereunderweight of which32.9% were severelyunderweight.

Also 27.3% ofchildren were wastedof which 4.7% wereseverely wasted.

There was a reduction inthe prevalence ofunderweight in childrenfrom 66.7% to 59.6%and severe underweightfrom 32.9% to 26.1%.

No confidence intervalsor p values were given

Successfulmanagement ofseveremalnutritionamongst childrenby workers atNCCs and infamily settingsusing standardprotocol led tothe wide scalereplication of theapproach byAnganwadicentres indifferent districtsof Rajasthan.

Ciliberto etal (2005)

To compare recoveryrates among childrenwith moderate andsevere wasting,kwashiorkor, or bothreceiving eitherhome-based therapywith RUTF orstandard inpatienttherapy.(Not clear from paperif emergency or not soit is assumed that thestudy period (2002 to2003) was during anemergency. Also, theOffice of ForeignDisaster Assistance ofthe Bureau forDemocracy, Conflictand HumanitarianAssistance supportedthe study).

Malawi (Dec 2002-June2003)

- A controlled,comparative, clinicaleffectiveness trial. Astepped wedge designwas used tosystematically allocatechildren into standardtherapy or home-based therapy withRUTF.


All children aged 10 to60mo in 1 of 7 NRUs andeligible for second phasetreatment for childhoodmalnutrition as well aschildren brought in fromthe community withwasting (WHZ < -2), mildoedema, or both with agood appetite were eligiblefor inclusion.

Of note, anthropometriccriteria for admission tothis study were those usedlocally in Malawi ratherthan those outlined in theWHO guidelines, whichdesignate that onlychildren with a WHZ < -3be offered standardinpatient therapy.

N = 1178 of which home-based was 992 andstandard therapy was 186.

Weight-for-height zscore > -2

Weight gain

Prevalence of fever,cough and diarrhoea

Children who receivedhome-based therapywith RUTF were morelikely to achieve a WHZ> -2 than those whoreceived standardtherapy (79% comparedwith 46%; p< 0.001)and were less likely torelapse or die (8.7%compared with 16.7%;p<0.001).

Home-base with RUTFgroup had greater ratesof weight gain (3.5compared with 2.0 g.kg-1.d-1.; difference:1.5; 95%CI: 1.0, 2.0 g.kg-1.d-1) and alower prevalence offever, cough anddiarrhoea than thestandard therapy group.

Home-basedtherapy withRUTF is associatedwith betteroutcomes forchildhoodmalnutrition thanis standardtherapy.

Rossi et al(2006)

To evaluate healthprojects andnutrition & foodsecurity programmeson nutritional statusin chronic post-conflict situation: thecase of Congo as anexample

Seven out of 14nutrition/foodsecurity projectswere visited inthe easternregions of Congofor a period of 4weeks (July toAug 2004).

The nutritionalapproach inCongo is theclassicalemergencyprogramme withNutrition Centres;TFCs admittingseverelymalnourishedsubjects in localhospitals andSFCs correctingmoderatewasting.

- Semi structuredindividual and groupwere held with keyinformants at the levelof implementingpartners (IPs), relevantMinistries, UNagencies, localauthorities, as well asrandomly selectedmembers of the targetpopulation groups.Interviews were oftencoupled with directobservation duringvisits to project sites.

Mortality and Morbidity statisticswere collected fromhospitals and healthcentres, whilenutrition and foodsecurity informationwere gathered fromdatabases provided bydifferent IPs.

- Statistics notpresented eg samplesize not stated.

- Children aged < 5y

- Older children (6-18yrs)

- Women of childbearingage

- Nutritional programmeswere evaluated accordingto internationally acceptedSphere standards.

- Acute malnutritionor wasting defined asWHZ < -2 using NCHSreference.

- Sphere keyindicators

Baseline Info- There was nobaseline information

The impact of nutritioncentres wasdemonstrated by thereduction of globalwasting and its severeform in the areascovered by nutritionalprogrammes. (Bar graphwas used to illustrate.CIs and p values werenot presented)

- Key indicators underthe sphere minimumcut-off points were goodfor 4 differentgeographical sitesValues were shown in atable.

- Of note, the presenceof patients in NCs wasinfluenced by security,possibility of access andseasonality.

The currenttargeted shortterm actionshould evolvetowards‘integrated basicneeds projects’,implemented byone or aconsortium ofpartners.

Also, health,nutrition andfood securitycomponentsshould beconsidered astandard publichealthinterventionstrategyrepresenting themost sensibleapproach toaddress the needsof the affectedpopulation.

Cont’d next page


39

Author ofstudy

Objective ofStudy


Surveypopulation

Indicator /outcome

Impact Conclusion

Collins et al(2006)

A review to assessthe impact ofcommunity-basedtherapeutic care(CTC) programs onsevere acutemalnutrition inmostly emergencysituations and a few(approx 2) non-emergencysituation

Malawi,Ethiopia,North Sudanand SouthSudan (2000-2005)

Compilation and analysesof CTC programs indifferent locations

Twenty one CTCprogrammeswere analysed.Severelymalnourishedchildren(definedaccording tothe criteria of<70% ofweight-for-heightpercentage ofmedian (WHM)or MUAC <110mm), aged< 5y.

N=23,511

- Recovery rates

- Weight gain

- Coverage rate

- Cost-effectiveness(data available for3 emergency CTCprogrammes)

Baseline infoNot readilyinterpretablebecause they areall in tabularforms.

Recovery rates of 79.4% wereachieved.

Rates of weight gain were between 4& 5g/kg/day which are < Spherestandards but did not result inincreased mortality or default rates.

Mortality rates of 4.1% were achievedwhich is < half of the Spherestandards. Default rates = 11%Transfer and non-recovery rates were3.3% and 2.2% respectively. Coveragerates were approximately 73%.

Of the severely malnourished childrenpresented, 76% were treated solely asoutpatients.

Cost effectiveness varied from US$12to US$132 per year of life gained.

Initial data indicatethat CTC programsare affordable, withthe cost-effectivenessof emergencycommunity-basedtherapeutic programsvarying from US$12to US$132 per year oflife gained.

Chaiken et al(2006)

To compare theeffectiveness of theCTC strategy incombination withconventionaltreatments foracute malnutritionat the individuallevel in droughtravaged ruralSouthern Ethiopia(An area which is notusually associatedwith the droughtcycle for whichEthiopia is known). This emergency wasidentified through aregional earlywarning system ofthe DisasterPrevention andPreparednessCommission.

SouthernEthiopia (Sept 2003-March 2004)

Data from monitoring theinitial phase ofprogrammeimplementation werereviewed to ascertainprogramme impact

Children agedless than 5ywith W/H <70% of normal,MUAC readingunder 11.0cm,or bilateraloedema intheirextremitieswere enrolledin the OTP. N= 5,799

Children withW/H between70% and 80%of normal wereenrolled in thesupplementalfeedingprogramme. N=7,961

- Under-5mortality rates

- Rates of severe acutemalnutrition SAM)

- Recovery ratesdetermined bygraduation to SFP.

- Default ratesand mortalityrates

- Coverage rates

Baseline infoAt start ofprogramme, dailyunder-fivemortality was1.47/10,000 andthe rate of SAMwas 1.0% (95%CI,0.5-2.0)

By the end of the programmemortality rates had improved to0.45/10,000 and SAM rates to 0.6%(95%CI, 0.2-0.9), respectively.

Out of the 5,799 children with SAM,over a 5-month period, 64.9%progressed sufficiently to graduate tothe supplementary feedingprogramme.

Default rate was low (2.3%),indicating community acceptance ofthe approach.

Mortality rate was low (0.2%)

OTP coverage was 78.3% and SFP was86.8% which are both excellent.

CTC is an importanttool to effectivelyaddress nutritionalemergencies and maybe a valuable entrypoint for long termdevelopment.

CTC may help makethe transition fromrelief todevelopment.

Greco et al(2006)

To assess the effectof milk plusporridge made fromlocal ingredientsversus milk aloneon nutritionalstatus during a warcrisis in NorthernUganda as milk isnot alwaysavailable outsidethe hospital andgiven thedisappointingresults ofnutritionalrehabilitation withmilk only.

St Mary’sNutritionalUnit in Lacor,NorthernUganda (2001-2003)

- Cohort study design inwhich 3 different cohortsof children were studiedbefore and after theintervention.

InterventionIntroduction of porridge inthe diet (called nutricam)in 2002.

Treatment periodAn average of 20 days

Cohort 1-before intervention, milkonly in 2001

Cohort 2- soon after intervention in2002Cohort 3- more than oneyear after intervention in2003

Children were randomlysampled for 3 months(Oct, Nov, Dec) for eachyear (2001, 2002, 2003) tocontrol for seasonal effects

- No control group


Children aged6mo to 6y withseveremalnutritiondefined asweight < 70%of thatexpected forlength orheight(marasmiccases),peripheraloedema scorebetween 1 and4+(kwashiorkorcases) or bothmarasmus andoedema.

N= 100 casesfor each yeargroup.

- Oedema freeweight gain

- Survival

- Cost

Baseline infoNone provided

* Of note, duringthe study, thenumber ofadmissionincreased and thestatus of thechildren atadmission wasworse, probablybecause warringfactions hadcaused manyfamilies toabandon theirvillage

Nutricam plus milk was more effectivethan milk alone in nutritionalrehabilitation of severelymalnourished children. Oedemadisappeared rapidly, and daily weightincrements rose significantlycompared with treatment with milkonly

- Average oedema-free weight gainincreased from 21 g/d (95%CI, 12-29)in 2001 to 35 g/d (95%CI, 25-45) in2002 and reached 59 g/d (95%CI, 51-65) in 2003.

- Mortality decreased from 22% to7.8%, and nutritional failures(insufficient weight gain) decreasedby greater than 50%.

- Nutricam is locally feasible at a lowcost (about €0.056/serving, includinglabour and fuel).

- The low-costporridge supplement(€2640/yr per 100children) waseffective in treatingmalnutrition.

- Widespread use ofthe porridge, whichresulted in betteroutcomes than milkalone, could producesavings in themedium to longterm, therebyreleasing resourcesfor other uses.

LimitationsThere couldn’t be acomparison of anuntreated cohortwith a differenttreated cohort andthe cohorts could notbe divided into milk-only andmilk-plus-nutricamsubgroups becausethe children wereseverelymalnourished and inneed ofsupplementation.

Cont’d next page


40

Author ofstudy

Objective ofStudy


Surveypopulation

Indicator /outcome

Impact Conclusion

Gaboulandet al (2007)

To assess the effectof three nutritionalstrategies onrecovery rates ofseverelymalnourishedchildren in Maradi,Niger(It is assumed thatthe study period(2002 to 2003) wasduring anemergency).

Maradi, Niger (August2002-October2003)

Cohort study made up of 3groups.

Group 1 maintained in aTFC during entiretreatment.

Group 2 initially treated atTFC and completedtreatment at home.

Group 3 were exclusivelytreated at home.

Children aged 6to 59 mo withseveremalnutritiondefined as W/Hindex < -3 z-scores orbilateral pittingoedema or MUAC< 110mm.

N= 1937 ofwhich 660 werein group 1, 937 ingroup 2 and 340in group 3.

- Weight gain

- Length of stayRecovery

- Case fatality

- Defaulting

Baseline infoNo baselineinformation provided.Also CIs and p valueswere not provided

All 3 cohort’s average weightgain and time in programmemet the internationalstandards (> 8g/kg/day, 30-40 days).

Default rates were 28.1, 16.8and 5.6% for TFC only, TFC plushome-based and home-basedalone strategies respectively.

The case fatality rates were18.8% for TFC only and 1.7%for home-based alone. Nodeaths were recorded in theTFC plus home-based group.

This study suggeststhat satisfactoryresults for thetreatment of severemalnutrition can beachieved using acombination of homeand hospital-basedstrategies.

Sadler et al(2007)

To compareprogrammecoverage of twotherapeutic feedinginterventions (CTCand TFC) forseverelymalnourishedchildren inneighbouringdistricts in Malawias a result of anational nutritionalemergency declaredby the MalawiGovernment in Feb2002.

Dowa district(CTCprogramme)and Mchinjidistrict (TFCprogramme)in Malawi (2002-2003)

- Observational study(two surveys implementedsimultaneously one ineach of the twoprogramme areas.)

- Thirty 100 km2 quadratswere sampled.Communities locatedclosest to the centre ofeach quadrat weresampled using a case-finding approach.

- Not a controlled study


Cases weredefined aschildren aged<5y with </=70% of W/Hmedian usingNCHS referenceor presence ofbilateral pittingoedema.

N= not stated

- Seven monthsintoimplementationof both studies, astudy wasconducted tocomparecoverage.

Baseline infoNot provided

Receipt of treatmentwas ascertained by thechild’s presence in a TFPor by documentaryevidence.

Coverage in eachquadrat was estimatedin two ways, a periodestimate that providesan estimation ofcoverage for the recentperiod preceding thesurvey and a pointestimate that providesan estimation ofcoverage at the exactpoint in time of thesurvey.

- Cost

Overall the period coveragewas 24.55% (95% CI, 17.8-31.4%) in the TFC programmeand 73.64% (95% CI, 66.0-81.3%) in the CTC programme.

The point coverage was20.04% (95% CI, 13.8-26.3%)in the TFP programme and59.95% (95% CI, 51.4 - 68.5%)in the CTC programme.

Cost of the CTC programmewas US$197 per child duringthe emergency set upprogramme.The equivalent cost data fromthe TFC programme was notavailable for comparison butlittle information fromelsewhere indicates that TFCcare can cost anythingbetween US$156 & US$355 perchild treated.

CTC gavesubstantially higherprogramme coveragethan a TFCprogramme.

Given effectivetreatment, thisenabled higherimpact of CTC onsevere malnutritionin this population.

Linneman etal (2007)

To determine theoperationaleffectiveness ofhome-basedtherapy with RUTFfor moderate andsevere childhoodmalnutrition insouthern Malawi

(Home-basedtherapy with RUTFfor the treatment ofmalnutrition hasbetter outcomes inthe research settingthan standardtherapy. Malawi is achronic food insecurearea with high ratesof childmalnutrition).

Twelve ruralcentres insouthernMalawi(May 2005 –May 2006)

ObservationalIntervention study

Intervention - 733 kJ kg/day of RUTF andfollowed fortnightly for upto 8 weeks.

- Staff at each centrefollowed one of threemodels: medicalprofessionalsadministered treatment(5 centres), patients werereferred by medicalprofessionals and treatedby community health aids(4 centres), or communityhealth aids administeredtreatment (3 centres).

- Regression modellingwas conducted todetermine what aspects ofthe centre (formal trainingof staff or location along amain road) contributed tothe outcome.

- No control group


Severelymalnourishedchildren (< 70%of reference W/Hor presence ofoedema) andmoderatelymalnourished(70-85% ofreference W/Husing WFPcategorization)with a goodappetite.

N= 2131 SAMandN= 806 MAM

- Weight gain

- Recovered (Ifchildren’s oedemaresolved and theygained weight suchthat their final W/Hwas > 85% of theirideal W/H),

- Failed (if they did notachieve a W/H > 85%of their ideal W/H orrelapsed requiringinpatient treatment),

- Died or dropped out.

- Aspects of the centre(formal training of staffor location along amain road)contributing tooutcome

Baseline info98% of children withSAM had kwashiorkor.Mean (SD) of WHZ was -1.5(1.4) in childrenwith SAM and -2.7(0.56) in childrenwith MAM respectively.

On average, children withsevere malnutrition gained 3.5g kg/day (SD= 4.1) andchildren with moderatemalnutrition gained 4.6 gkg/day (SD= 4.1)

Recovery was 89% in childrenwith SAM and 85% in childrenwith MAM.

Thirty-four (4%) of themoderately malnourishedchildren failed, with 20 (2%)deaths, and 61 (3%) of theseverely malnourishedchildren failed, with 29 (1%)deaths.

Centre location along a roadwas associated with a pooroutcome.

Outcomes for severelymalnourished children wereacceptable with respect toboth the Sphere guidelinesand the Prudhon case fatalityindex.

Home-based therapywith RUTF yieldsacceptable resultswithout requiringformally medicallytrained personnel;furtherimplementation incomparable settingsshould be considered.

Cont’d next page


41

Author ofstudy

Objective ofStudy


Surveypopulation

Indicator /outcome

Impact Conclusion

BelachewandNekatibeb(2007)

To document theexperiences andlessons for rollingout of the OTPservice at thewider scale withthe aim ofassessing thestrengths andweaknesses ofthe project andsuggestrecommendationsfor futureprogramming.

(Ethiopia is anarea with chronicfood insecurityand nutritionalproblems causedby successivedroughts).

OTP pilotprogrammeimplementedin 3 regionsnamely;SouthNations andNationalitiesPeoplesRegions(SPNNR),Addis Ababaand Oromiaregions ofEthiopia(16-25 Nov2006)

Qualitative study

- Qualitativemethods of datacollectionincluding focusgroup discussion,observation andin-depthinterview of keyinformants.

- Review ofhealth facility,reports andprogrammedocuments.

Subjects included36 key informantsand 30 focusgroup discussants

- Attitude change

- Utilisation of existinghealth equipments

- Awareness ofmalnutrition and itstreatment

- Reduction in theburden of malnutritionand associated mortality

OTP has enhanced community’sunderstanding of malnutrition as ahealth problem through anexcellent entry point it created forBCC on optimal IYCF.

It has also enhanced utilisation ofthe existing equipments of therespective health services topromote nutrition and increasedmental satisfaction of the providerswho observed rapid recovery ofmalnourished children taking theplumpy nut.

OTP resulted in increased awarenessof the community aboutmalnutrition and its treatment,resulting in increased need-baseddemand for the OTP and self-referralof children to health facilities.

Shift in the thinking of providers onthe fact that malnutrition can betreated without admitting the childand reduction in the burden ofmalnutrition and associatedmortality are other positive findingsof the study.

While it was observedthat the programmewas very effective intreating case of SAMand is highly acceptableby planners, health careproviders andbeneficiaries, therewere differentoperational issues thatneeded to bestrengthened.

The irregularity andincompleteness ofsupply availability, highattrition of trainedhuman power,inadequate supportivesupervision especiallyfrom local MOH,inadequate communitymobilisations are someof the shortcomingsidentified. Based onthese findingsrecommendations wereforwarded.

Rossi et al(2008)

To evaluate theimpact andappropriatenessof programmesfor themanagement andtreatment ofseveremalnutrition inan emergencysituation inBurundi.

Nutritioncentres inBurundi (1999 –2004)

- But mostlyspecific on2004 data

Impact of previousnutritionprogrammes

Longitudinal datacollected in across-sectionalpopulation-basedsurvey

This evaluationNationalestimates usingcluster surveysand standardisedmethodology fordata collection

All age groups

From 1999-2004N= 1,054,210severelymalnourishedpatients of which131,668 weretreated in TFC and922,542 in SFC.

For 2004 onlyTwenty TFCs and224 SFCs wereactive in BurundiN= 127,420patients of which15309 weretreated in TFC and112,111 in SFC.Children < 5yaccounted for62% ofbeneficiaries inTFC and 76% inSFC.

- Under 5s mortalityrates

- Acute malnutrition

- Coverage

- Recovery rate, %

- Defaulter rate, %

Baseline infoIn 2000, under-5smortality rate wasestimated at 6deaths/10,000/d whileprevalence of acutemalnutrition was 8% andits severe form was 0.5%

No CIs or p values werereported

- The most convincing impact of thenutritional programme in Burundiwas the reduction of mortality ratein children <5y to rates of between3.1 to 4.9 deaths/ 10000/d.

- There was a small reduction onthe rates of acute malnutrition to7% and a worsening of theprevalence of severe forms to 1.1%

- CoverageNutritional programmes werepresent in every province with acoverage rate of 55%.

- TFC performance indicators(results shown as a bar graph)fulfilled the minimum standards indisaster response while SFC had alow recovery rate (69% v >80%) anda high non- respondent rate (10% v< 5%).

- Combining coverage and curerates, the programme met 44% ofthe assessed needs in 2004.

An impact on theprevalence of acutemalnutrition could notbe demonstrated.

In Burundi, thestabilisation of securityconditions permitted acombination ofhumanitarian responsesranging fromemergency activities tostrengthening ofcommunity-basedinitiatives that couldcorrect the coverage andimpact limitations.

Dubray et al(2008)

To compare theeffectiveness oftwo systemicantimicrobialtreatmentsadministered tochildren with SAMon admission in aTFC in an IDP areain south Sudan.

Mayo TFC,Khartoum-Sudan. (Jan2002-Sept2003).

- A randomised,unblended,superiority-controlled trialcomparing: oncedaily injection ofintramuscular(IM) ceftriaxonefor 2 days vs twicedaily oralamoxicillin over 5days (10 doses).

- No controlgroup

- Nutritionrehabilitationtook place in 3phases;stabilization,rehabilitation andadministration ofdrugs

From aninternallydisplacedpopulation,children aged 6 to59 mo with SAM(defined as W-H%< 70% of thereference medianusing NCHS/CDC1977 growthreference curves)and/or presenceof oedema and/ofMUAC < 110mm.

N= 458 of which230 = amoxicillingroup and 228 inthe ceftriaxonegroup.

- Weight gain (WG) (Treatment was definedas successful whenchildren had gained ≥10g/kg/day by the 14th dayof WG).

- Recovery rate

- Case fatality ratio (CFR)

- Defaulter rate

- Referral rate

Baseline infoIn the amoxicillin group,72.1% had W-H% < 70%without oedema, 15.7%had MUAC < 110mmwithout oedema and12.2% had bilateraloedema.In the ceftriaxone group,it was 74.1%, 15.8% and10.1% respectively.

In the intension to treat analysis,53.5% (123/230) in the amoxicillingroup and 55.7% (127/228) in theceftriaxone group (difference 2.2%95%CI, -6.9 to 11.3) had a WG >/=10 g/kg/day during a 14-day period.

Recovery rates were 70% and 74.6%in the amoxicillin and ceftriaxonegroups respectively with CFR being3.9% and 3.1% respectively.

In the absence of severecomplications, eitherceftriaxone oramoxicillin isappropriate formalnourished children.

However in ambulatoryprogrammes, especiallywhere there are largenumbers, ceftriaxoneshould facilitate thework of medicalpersonnel.

Cont’d next page


42

Author ofstudy

Objective ofStudy


Survey population Indicator /outcome

Impact Conclusion

Colombattiet al (2008)

To determine (i) the extent ofmalnutrition andthe risk factors forseveremalnutrition inGuinea Bissau, apost-conflictcountryexperiencing longtermconsequences ofcivil war and (ii) the feasibilityand effectivenessof a short-terminterventioncharacterized byoutpatienttreatment withlocally producedfood for thetreatment ofseveremalnutritionduring the rainyseason.

Outpatientclinic inBissau,GuineaBissau (July1- Aug 12,2003)

Intervention study

Intervention= localsemi porridge whichconsisted of blackmillet, wheat, drybanana, cabaceira,powdered milk, dryegg yolk, sugar, saltand a vitamin mix. Itwas cooked every dayin water and oil wasadded to reach anenergy density of1808 kJ/100 g (432kcal/100 g).

- No control group


N= 2642 children, age range:1mo -17y. Severelymalnourished children were

Results for objective (i)2272 (85.9%) of the examinedchildren had a certain degree ofmalnutrition (shown as a bargraph) Weight-for-age was < -2SD in23% of the children and <-3SDin 10.3%. 1.4% were severelymalnourished (defined as W/A <-3 SD using NCHS/ WHOreference) and admitted for daycare [nutritional outpatient unit(NOU]. - N=37

Significant differences betweenthe general population and theseverely malnourished childrenobserved were severelymalnourished children in poorhealth status were mainlyinfants (P<0.001), part of largefamilies (P<0.001), did not havea bathroom at home (P<0.001),did not use aqueduct water(P<0.001) and lived in mudhouses (P<0.001). Mother’silliteracy was higher in themalnourished group (P<0.001).

Weight gain

Length of stay

Cost

All 37 children recoveredwith a weight gain of 4.45g/kg/d (range: 0.97-8.67g/kg/d, with an overallweight increase that rangedfrom 1% to 25% of initialbody weight.

Mean length of stay in NOUwas 19d (range: 2-33d).

At 1y follow up, 36 childrenwere alive, had maintainedweight gain (with W/H of 0to +1SD and W/A of 0 to+1SD) and avoided relapse.One child died from HIV-related TB.

CostsMean cost of €5 per child.Overall cost of theintervention was €13448 ofwhich foreign personnelcost €6219, local personnelsalaries, medical supplies,drugs for 2642 children,health education materialsfor their mothers cost€7229.

Short-terminterventionsperformed inpost-conflictcountries duringseasons of highburden ofdisease andmalnutrition, isfeasible andsuccessful at lowcost.

Day-caretreatment ofseveremalnutritionwith locallyproduced food isan option thatcan be tested inother settings.

Amthor et al(2009)

To describe theimpact of OTP inaddressing theoperationalimpasse ofgeographicdistance and lackof existing healthcentre support intreating childhoodmalnutrition in arural area during afood crises.This was as a resultof the faminedeclared in Malawiin Jan of 2006 bythe internationalcommunity.

Machingadistrict,Malawi(March –July 2006)

- Intervention studyusing RUTF for 8 weeks

- No control group

Severely malnourished childrenaged 6 to 60 mo with anadequate appetite presented to1 of 5 OTP centres in Machingadistrict. N=826

(Severe malnutrition defined aseither the presence of oedema orhaving a W/H < 70% of the 2006WHO reference standard.

- Recovered, definedas reaching 85% ormore of ideal weight-for-length

- Remainedmalnourished

- Defaulted

- Died

Baseline infoNot provided

Out of 826 children 93.7%recovered, 1.8% remainedmalnourished, 3.6%defaulted and 0.9% died.

The fraction of children withsevere malnutrition beforeand after treatment wassignificantly less, p<0.0001using x2 test.

Mean weight gain was2.7+/-3.7 g/kg/d, heightgained 0.3+/-0.9mm/d, andMUAC gained 0.2 ±0.3mm/d.

Using a paired t-test, allthese outcomes weresignificantly different aftertreatment (p<0.0001).

Home-basedtherapy withRUTFadministered byvillage healthaides is aneffectiveapproach totreatingmalnutritionduring foodcrises in areaslacking healthservices.

Guerrero etal (2010)

To identify theprinciple factorsthat negativelyaffects thecoverage of CTCprogrammes.(Data are mostlyfrom emergencysituations)

the DRC,Ethiopia,Malawi,Niger andNorth andSouthSudan(2003-2006)

Retrospective analysisof quantitative &qualitative data.

Quantitative data wasfrom Centric SystemArea Sampling (CSAS)Coverage Surveys.

Qualitative data wasprocured throughsocio-culturalassessments of theseprogrammes.

- Questionnaireadministered to carersof cases of malnutritionwho were not enrolledin the programme hasbeen used toinvestigate the reasonsfor non-attendanceand to pinpoint issuesunderminingprogramme coverage

Caretakers of childrenN= 1,696

Reasons for non-attendance to CTCprogrammes- Fear of rejection- Distance to sites- Lack of awarenessof the programme- Insecurity on route- Negative clientexperiences at sites- Condition notalways recognised asmalnutrition- Opportunity costs- Husband’s refusal- Palatability of food- Shame- Elderly-headedhousehold- No one to care forother children athome- Fear of cross-infection at the sites- Harvest obligations

- Previous rejection fromprogramme was identifiedas a reason for non-attendance by CSAS surveyquestionnaires in 10 of the12 programmes

- Awareness of theprogramme and theconditions that it treats, wasalso identified as a maindeterminant of coverage

- Distance to sites wasfound to be the primarybarrier to access for 10.8%of severely malnourishedcases not enrolled in CTCprogrammes.

- CTCprogrammecoverage inemergencies canbe improved bytackling threecommon factors;distance to sites,communityawareness ofthe programmeand the way inwhich rejectionsare handled atsite, whichtogetheraccount forapproximately75% of non-attendance.


43

Author ofstudy

Objective ofStudy


Surveypopulation

Indicator /outcome

Impact Conclusion

Lopriore et al(2004)

To assess the effectof a highlynutrient-densespread fortifiedwith vitamins andminerals, and/orantiparasiticmetronidazoletreatment, incorrecting retardedlinear growth andreducing anaemiain stuntedSaharawi refugeechildren. (After politicalchanges during the1970s, theSaharawi peoplefled their homelandin the WesternSahara desert andfor more than 25yhave lived in whatcan be defined as apermanent state ofemergency).

Saharawirefugeecamps, westAlgeria (May1998- Jan1999)

- Randomised,double- blind,placebo-controlledsupplementationtrial with orwithout anti-parasitictreatment. Childrenwere assigned to 1of 5 groups namely;Fortified spread(FS),

Fortified spreadand metronidazole(FS+M),

Unfortifiedspread(US),

Unfortified spreadplus metronidazole(US+M),

Or control.

- Supervisedsupplementationwas given daily for6 months

- Indicators wereassessed at 0, 3 and6 months.

- Statisticspresented

Of 1144 childrenidentified in thecensus conductedin the refugeecamps, 374children wereeligible to enterthe study.Stunted childrenaged 3-6yN= 374, a total of75 children foreach group (Stunting definedas HA < -2 zscores usingWHO/NCHSreference).Intervention –50g/d of a fatspread high inenergy & proteinsand fortified withvitamins andminerals.

- Linear growth

- Increase in Hb concn.

- Anaemia

Baseline info254 children completedthe study

FS groupN= 51HAZ =-2.87± 0.56Hb(g/L)= 92 ± 23% with anaemia= 70.6%

FS + M groupN= 52HAZ =-2.8± 0.62Hb(g/L)= 90 ± 22% with anaemia= 82.7%

US groupN= 52HAZ =-2.73± 0.55Hb(g/L)= 91 ± 19% with anaemia= 82.7%

US + M groupN= 54HAZ =-2.97± 0.84Hb(g/L)= 89 ± 22% with anaemia= 83.3%

Control groupN= 45HAZ =-3.00± 0.71Hb(g/L)= 91 ± 23% with anaemia= 82.2%

- Linear growth of children fed FSwas 30% faster at 3mo than in US andcontrol groups, after which HAZincreased only slightly in the FS groupand remained unchanged in the othergroups.

At 6 mo, HAZ was significantlyincreased in the FS groups (0.15 ±0.22), whereas it remainedunchanged in the other groups(MANOVA, P < 0.0001). Themagnitude of the catch-up growthresponse was negatively associatedwith the age at baseline (β = -0.034;95% CI:-0.045, -0.022); hence, theyounger the age, the greater thegrowth increase. This effect remainedafter correcting for differences intreatment between groups, initial Hbconcn, and baseline HAZ (β = -0.0315;95% CI:-0.043, -0.020). Baseline HAZwas negatively associated withgrowth increase, but the effect wasnot significant (β = -0.097; 95% CI: -0.289, 0.096).

- No additional benefits frommetronidazole were observed.

- Increase in Hb conc. in the FS groupat 6mo was twofold that in the USand control groups (37±40, 19±15and 16 ± 17 g/L respectively;p<0.0001) and anaemia was reducedby nearly 90%.

- FS, and not US,induces catch-upgrowth in stuntedchildren whosediets are poor inmicronutrients.

- This trial providessupport fordelivering multiplemicronutrients toreverse stuntingand reduceanaemia in childrenup to age 6y.

Moench-Pfanner et al(2005)

To assess the effectof food-for-workprograms (a postcrises response) onnutrition outcomesparticularlyanaemia.(Indonesia’seconomic crisis oflate 1990’s loweredconsumption ofmicronutrient-richfoods, whichincreased theprevalence ofmicronutrientdeficiencies,includinganaemia).

Food-for-work (FFW)wereimplementedin 3 urbanand 2 ruralsites inIndonesia(Dec 2000-Sep 2003)

- A quasi-experimentaldesign in whichboth programmebeneficiaries aswell as non-participatingcomparisonhouseholds(controls) werefollowedlongitudinally.

- Control group

- Surveys wereconducted atbaseline and at6mo intervals for2.5y


Children aged andmothers

N= 1500randomlyselected HH and1500 controls ineach site usingintervalsampling.

- Mainly anaemia

- To better understandthe “income effect” ofthe FFW programs, HHexpenditure datawere collected by askingthe respondents for theamount spent in theprevious week on >20items and on categoriesof expenditure.

Baseline info(Provided but in tabularform)

Food-for-work programs had limitedeffect on anaemia.

Only among urban poor mothers inSubaraya were the odds of anaemiaat endline lower when participatingin the FFW programme (0.6, 95% CI[0.40-0.89]).

The poor were found to beappropriately targeted andprogramme participation rangedfrom 4 to 18 months.

The proportion of households withdebts ranged from 32 to 70%;although it was higher amongbeneficiaries than controls, itincreased among controls, but notbeneficiaries.

Micronutrientdeficiencies shouldbe addresseddirectly viasupplements andfortified foods.Closer attention is

required to thepotential foraffectingnutritionaloutcomes throughFFW, including foodaid quality andquantity andcomplementarynon foodinterventions.

Cont’d next page

4. Summary of published studies assessing the impact of micronutrients distribution programmes in emergency situations


44

Cont’d next page

Authorof study

Objective ofStudy


Surveypopulation


Nielsen etal (2005)

To evaluate theimpact of vit Asupplementationon childhoodmortality duringthe war -emergency inGuinea- Bissaufrom June 1998– May 1999.Vit Asupplementationis recommendedby WHO inemergencysituations but noreported study onthe impact of VitA in emergencieshave been found.

Bissau,GuineaBissau (Oct1, 1998-Mar 31,1999)

Vit Asuppleme-ntation hadnot beenprovidedroutinely inGuinea-Bissau inpeacetime.

- Observational study.Since study couldn’t berandomised, impact wasevaluated in different ways;

- Used variation in thedelay of provision of vit A ina step-wedged design

- Compared wartime withpre-wartime mortality

- Examined whether vit Aas a free commodityreduced cultural and socialeconomic inequalities inchildhood mortality.

- No control group


Pre-war time (Oct 1, 1995-Mar 31, 1996 + Oct 1, 1996-Mar 31, 1997 + Oct 1, 1997-Mar 31, 1998)

Children aged 6-59 mo living inBissau.

N= 5926 of which2911 received atleast one dose ofvit Asupplementation.Children aged 6-12 mo received atleast one oraldose of 100,000IU while olderchildren received200,000 at homeevery threemonths by a fieldworker.

Mortality rates (deaths/ PY)

Baseline infoPre-warMortality rates (deaths/ PY) =0.02 (76/3399)

Using the variation in theprovision of Vitamin A, a slightnon-significant reduction inmortality for children between 6months and 5 years of age(mortality ratio (MR) 0.49; 95% CI0.09–2.70) was found.

Comparing with a three-yearperiod before the war, childrenoffered Vitamin A at home duringthe war had a 12% reduction inmortality (MR 0.88; 0.41–1.87),whereas the overall impact of thewar was an 89% increase inmortality (MR 1.89; 1.32–2.71).

Vitamin A supplementation wasassociated with a reduction incultural and socio-economicinequalities.

Vit Asupplement-ation mayhave abeneficialimpact onchildhoodmortality in anemergencysituation`.

Menon et al(2007)

To assess theeffectiveness ofan interventionproviding 2 moof micronutrientSprinkles(containing 12.5mg iron andother keymicronutrients)in reducinganaemia amongchildren in ruralHaiti. (An areawhere more thanone-half ofchildren underthe age of 2y areanaemic).

CentralPlateauregion ofHaiti(Mar-Sep2005)

A cluster randomized pre-post intervention designthat compared before- andafter-Hb measurementsand prevalence of anaemiaamong children whoreceived Sprinkles alongwith their Wheat-soy blendration (S-WSB group) andchildren who received onlyWSB (WSB group).

Sprinkles interventionincluded 2 parts: a 2-mosupply of Sprinkles and aBCC intervention, whichconveyed information onappropriate use of theSprinkles. There was a 2moand 7mo follow up period.

- Untreated control groupfor 2 months


Children aged 9to 24 mo with Hb> 70g/LN= 415 of which254 in S-WSBgroup and 161 inWSB group

- Hb levels

- Anaemia prevalence (Hb < 100g/L)

- Cost

Baseline infoAt baseline, anaemia prevalenceadjusted for age and sex, was54% and 39% in S-WSB and WSBgroups, respectively.

- After the 2-mo intervention (1stfollow-up), anaemia, adjusted forbaseline prevalence, age, and sexdropped to 24% in S-WSB (P<0.001) and increased to 43% inWSB (P=0.07).

- At 7mo post intervention,anaemia in S-WSB declined to 14%;92% of children who were nonanaemic at 1st follow-up remainedso without further Sprinklesconsumption. From baseline to 1stfollow-up, mean Hb increased by5.5 g/L and dropped by 1.0 g/L inthe S-WSB and WSB groups,respectively (P<0.001). Frombaseline to 2nd follow-up, mean Hbincreased by 10.9 g/L in S-WSB(P<0.001). Changes in mean Hbwere greater for younger children(<21 mo at onset of intervention)(P<0.05) and for children who wereanaemic at baseline (P < 0.001).

The total cost of the 2-moSprinkles supply was $2US.

In populationswith a highprevalence ofanaemia, suchas rural Haiti,2 mo ofSprinkles areeffective inreducinganaemiaamong 9- to24-mo-oldchildren.

Seal et al(2008)

To assess theimpact offortified maizemeal on iron andvitamin A statusof food aiddependentrefugees

Nangweshirefugeecamp,Zambia(July 2003,July 2004)

- Pre- and post-intervention using alongitudinal cohort(subjects were followed upafter 12 months.

- No control group


- Children aged6-69 months.N= 157

- Adolescentsaged 10 -19yN= 212

- Women aged 20 - 49yN= 118VBN ,M

sTfR = Serumtransferrinreceptor

- Hb conc.

- Anaemia defined as Hb < 110g/l

- Serum transferrin receptor

- Vitamin A status

Baseline info(Means & 95%CI are shown)

Children n=155Hb(g/dl) = 10.9 (10.6, 11.2);Anaemia (%) = 47.7 (39.7, 55.9);WHZ = -0.14(-0.27, -0.01);HAZ = -1.83 (-2.01, -1.64).

Adolescents n=213Hb(g/dl) = 12.9 (12.7, 13.2);Anaemia (%) = 19.2 (14.2, 25.2);Log10 sTfR(μg/ml) = 0.875(0.849, 0.901);% Fe deficient (sTfR > 8.3μg/ml)=31.5(25.2, 38.4);Serum retinol (μmol/l) = 0.73(0.69, 0.77);% vit A deficient (serum retinol<0.7μmol/l) = 46.4 (39.4, 53.4).

Women n= 91Hb (g/dl) = 13.1 (12.7, 13.5);Anaemia (%) = 16.5 (9.8, 26.1).

During the intervention period,mean Hb increased in children(0.87g/dl; p<0.001) andadolescents (0.24g/dl; p=0.043)but did not increase in women.

Anaemia decreased in children by23.4% (p<0.001) but nosignificant change in adolescentsor women.

sTfR log10-transformed decreasedby -0.082μg/ml (p=0.036)indicating an improvement in theFe status of adolescents but therewas no significant decrease in theprevalence of deficiency (-8.5%;p=0.079).

In adolescents, serum retinolincreased by 0.16μmol/l(p<0.001) and vitamin Adeficiency decreased by 26.1%(p<0.001).

The introduc-tion offortified maizemeal led to adecrease inanaemia inchildren and adecrease invitamin Adeficiency inadolescents.Centralised ,camp-levelmilling andfortification ofmaize meal isa feasible andpertinentintervention infood aidoperation


45

Author ofstudy

Objective ofStudy


Surveypopulation


Magoni et al(2008)

To evaluate theeffectiveness ofan emergencyproject (universalironsupplementationand foodsupplementation)for the reductionof iron deficiencyand malnutritionamong childrenin Palestine dueto the emergencysituation causedby the outburst ofthe secondintifada inSeptember 2000.

- Yattaarea, WestBank(2002-2005)

- Southernand South-Westernareas ofHebrongovernorate, WestBank,Palestine(August2004 – July2005) laterreferred toas otherareas

- Two crosssectional surveysto measure beforeand afterintervention oftwo independentrandom samplesof children livingin target areas.

- No controlgroup


Before the projectSubjects aged 6–59months wererandomly selectedfrom a list of childrenliving in target areasN= 266 of which 152was Yatta area and114 other areas

After the projectSubjects aged 9–59months wererandomly selectedfrom a list of childrenliving in target areasN= 367 of which 150was Yatta area and217 other areas

- Information aboutuse of ironsupplementationduring project periodwas collected.

- Prevalence of anaemia (definedas Hb level < 11g/dl).

- Wasting or acute malnutrition,stunting or chronic malnutritionand underweight (defined as W/H,H/A and W/A < -2 SD of NCHSreference respectively).

Baseline infoPrevalence of anaemia,underweight and wasting for allchildren were 30.1% (CI 95%:24.5–35.6%), 10.9% and 6.0%respectively.

- Using geographical areasanaemia, wasting, stunting andunderweight were 17.1%, 5.9%,22.4% and 14.5% for Yattarespectively and 47.4%, 6.1%,11.4% and 6.1% for other areasrespectively.

The prevalence of anaemia for allchildren decreased from 30.1% (CI95%: 24.5–35.6%) to18.8% (CI95%: 14.8–22.8%; decrease of38%, p = 0.001). Levels andreduction in anaemia prevalencewere different according togeographical areas: in Yatta areawhere prevention activities hadbeen implemented in previousyears, prevalence of anaemia was17.1% and remained stable atlower levels (14.7%). In otherareas it sharply declined from 47.4to 21.7% (p < 0.0005): the declinewas present only for children whoreceived iron supplementation.

Comparison of anthropometricsindex before versus after the projectshowed that underweight declinedfrom 10.9 to 3.8% (p = 0.0006) andwasting declined from 6.0 to 1.4%after (p = 0.0025) for all children.

With simpleepidemiological tools we coulddemonstrateand measuretheeffectiveness ofourinterventionson the healthstatus of thegeneralpopulation: a50% reductionof anaemia anda 70%reduction ofglobal acutemalnutrition.

Talley et al(2010)

To evaluate theeffectiveness ofstainless steel (Fealloy) cookingpots in reducingFe-deficiencyanaemia in food-aid dependentpopulations

Two long-termrefugeecamps inwesternTanzania(Dec 2001-Jan 2003)

- Pre- and post-intervention(stainless steelpots) withrepeated cross-sectional surveysconducted at 0, 6and 12 monthspost intervention.

One camp servedas interventionwhile the otherserved as a controlgroup (iecontinued usingaluminium andclay pots).


Children aged 6 to 59monthsN=approx 110

Non pregnantmothersN= approx 110

sTfR = Serumtransferrin receptor

IQR= Interquartilerange (25th, 75thpercentiles).

- Hb conc.

- Anaemia (defined as Hb < 11.4g/dl and 12.4 g/dl (because of highaltitude) for children and nonpregnant mothers respectively)

Baseline info*= mean (SD) ˚= Median & IQR.

Intervention group*Hb (g/dl) = 11.5 (1.6) and 14.5(1.4) for children and mothersrespectively while ˚sTfR (μg/ml)=7.5 (6.0, 9.9) and 5.9 (4.8, 5.9).

Control group*Hb (g/dl) = 11.7 (1.4) and 13.7(1.5) for children and mothersrespectively while ˚sTfR (μg/ml)=8.4 (6.5, 10.8) and 5.7 (3.6, 7.3).

ChildrenThere was no change in Hb conc.at 1 y; however, Fe status waslower in the intervention campthan the control camp (sTfR conc:6.8 v 5.9 μg/ml; p<0.001).

Non pregnant mothersThere was no change in Hb conc.at 1 y; however, Fe status waslower in the intervention campthan the control camp (sTfR conc:5.8 v 4.7 μg/ml; p= 0.003).

Distribution ofstainless steelpots did notincrease Hbconc. orimprove Festatus inchildren ortheir mothers.

The distributionof stainlesssteel pots inrefugee contextis notrecommendedas a strategy tocontrol Fedeficiency.


46

5. Summary of published studies assessing the impact of food security or agricultural intervention in emergency situations

Author ofstudy

Objectiveof Study


Surveypopulation

Indicator /outcome

Impact Conclusion

Doocy et al(2005)

To assess theimpact of amicrofinanceprogrammeonnutritionalstatus andassessmentof copingcapacity inchronicdroughtconditions.A severedroughtaffected thestudy areasin2002/2003

Two ruralsites(Adamasite & Sodosite inEthiopia(May 2003)

- Observational study (Athree group cross-sectionalsurvey)

This study compared twogroups of clients thatreceived loans with a controlgroup.

i) established clients- havecompleted two or more loancyclesii) incoming clients- have

completed one loan cycle orless and been participatingin the programme for nomore than 10 mo) iii) Community controls- hadnot received a loan withinthe past year and were notseeking a loan.

The sample was stratified bysurvey site and client sex.


Total of 819households

Children aged6-59 months N = 608

FemalerespondentsN=352

MalerespondentsN=456

- Severe and moderateto severe acutemalnutrition usingMUAC of <11.0cm and <12.5cm respectively forchildren

- Acute malnutrition forfemales and males usingMUAC of < 22.0cm and<23.0cm respectively

Baseline infoNone provided

*Study limitationsi) non randomization ofsubjects to experimental& control groups

ii) Cross sectionaldesign’s Inability tocontrol for secular trendsor directional changes incharacteristics of thepopulation over longperiods of time.

In the total sample, 5.8% of children wereseverely wasted and 10.2% of children weremoderately wasted. The prevalence ofoedema was 1.8%, and 17.8% of childrenwere classified as having global acutemalnutrition.

In the primary survey site, Sodo, femaleclients and their children had significantlybetter nutritional status than othercomparison groups: the odds ofmalnourishment in female communitycontrols compared to established femaleclients was 3.2 (95% CI: 1.1–9.8) and theodds of acute malnutrition in children 6–59mo were1.6 times greater in children of bothmale clients and community controls (95% CI: .78–3.32).

Male clients and community controls were1.94 (95%CI:1.05-3.66) and 2.08 (95%CI:1.10- 4.00) times respectively more likelyto have received food aid during the past yearthan female clients

Findings of thestudy suggestthatmicrofinanceprograms mayhave animportantimpact on thenutritionalstatus andwell-being offemale clientsand theirfamiliesespecially inthe context ofdrought andfood insecurity.

Abebe et al(2008)

To describeanemergencylivestock-relatedintervention(commercialdestocking)in place offood aid in adroughtravaged area

Moyaledistrict,southernEthiopia(2006)

- Intervention study

Intervention = destocking.An estimated 20,000 cattlevalued at USD 1.01 millionwas purchased

- Impact was assessedusing both participatoryapproaches andconventional sampling withstatistical analysis


Pastoralistshouseholds,N=approx5,405 HH

For impactassessmentSeven kebeles& out of the570 HH thathad destocked,20% wererandomlyselected N= 114 HH.

(A kebele is acluster ofvillagesrepresentingthe smallestadministrativeunit inEthiopia).

- Saves human lives

- Helps fast recoveryand herd rebuilding

- Helps to cope witheffect of the drought

- Helps the livestock tosurvive

- Benefits the poor themost

- Socially and culturallyacceptable

- Timely and available

- Overall preference

Means and 95%CIprovided in graph forms

Impact on livelihoods: income andexpenditure

- On average, destocked households receivedUSD 186 from the sale of cattle

- Destocking was considered to be the mostuseful intervention with mean score 9.1 (95%CI:8.5, 9.7) than any otherintervention.

- Income from destocking accounted for54.2% of HH income (which was significantlyhigher than any other source at 95%confidence level) of which 79% was used tobuy food, care for livestock, meet variousdomestic expenses, support relatives, andeither pay off debts or augment savings.

- Expenditure on livestock care amounted to36.5% of local spending.

- Informants confirmed they were able tobuy their own food (as well as other thingslike medicine and clothes) with moneyobtained from destocking instead of havingto wait for food aid as in previous years.

- Food aid was perceived as the third mostpreferred option and was a particularlyimportant type of support for poorer HH

Cost-benefitThe cost-benefit ratio was 41:1

Commercialdestocking is aviable anduseful droughtintervention.

The potentialfor commercialdestocking toreach the mostvulnerablepastoralist HHrequires furtherresearch, asdoes the mostappropriatecombinationsof livelihoods-basedinterventionssuch asdestocking andfood aid.

Cont’d next page


47

Author ofstudy

Objective of Study Setting Study design, Statistics Surveypopulation

Indicator /outcome

Impact Conclusion

Bezner et al(2010)

To investigate whetherchildren in householdsinvolved in aparticipatory agricultureand nutritionintervention hadimproved growthcompared to children inmatched comparablehouseholds and whetherthe level of involvementand length of time in theproject had an effect onchild growth.

(An area with chronicfood insecurity, highlevels of childmalnutrition and highlevels of HIV prevalence)

A rural area innorthernMalawi(2001-2007)

A participatoryagriculture andnutrition project(the Soils, Foodand HealthyCommunities(SFHC) project)was initiated byEkwendeniHospital aimedat improvingchild nutritionalstatus withsmallholderfarmers

A prospective quasi-experimentalstudy

- Compared baseline and followup data in ‘intervention’ villageswith matched subjects in‘comparison’ villages. Mixed modelanalyses were conducted onstandardized child growth scores(weight- and height-for-age Z-scores), controlling for child ageand testing for effects of length oftime and intensity of villageinvolvement in the intervention.

- Agricultural interventionsinvolved intercropping legumesand visits from farmer researchers,while nutrition education involvedhome visits and group meetings.

Participants ininterventionvillages wereself-selected, andcontrolparticipants werematched by ageand householdfood securitystatus of thechild. Over a 6-year period, ninesurveys wereconducted,taking 3838height andweight measuresof children underthe age of 3 years

- Child growthpattern

Baseline infoControl andinterventionhouseholds showed nosignificant differencesin landholding size(average of 1.1 ha),formal education (86–87% with someprimary education),livestock ownership(72% v. 78% ofhouseholds) or childnutritional status(47% v. 48% ofchildren stunted).

There was animprovement overinitial conditions of upto 0.6 in WAZ (from -0.4 (SD 0.5) to 0.3 (SD0.4)) for children in thelongest involvedvillages, and animprovement overinitial conditions of 0.8in WAZ for children inthe most intenselyinvolved villages (from-0.6 (SD 0.4) to 0.2 (SD0.4)).

Long-termefforts toimprove childnutritionthroughparticipatoryagriculturalinterventionshad asignificanteffect on childgrowth.


48

6. Summary of published studies assessing the impact of general ration distribution programmes in emergency situations

Authorof study

Objective ofStudy


Indicator /outcome

Impact Conclusion

De Waal et al(2006)

To investigatewhether a vast,semi-nationwidedistribution ofgeneral rations,improved thesurvival chancesof children in thegeneralpopulation.

(Distribution ofgeneral rationswas in responseto the 2002/2003drought)

Rural andurban anddrought-affectedandunaffectedlocalities inEthiopia (Apr-Jun2004)

Observational study (cross-sectionalsurveys)Data were collected on patterns ofmortality, socio-economicdeterminants of child survival, andthe food, water, and livelihoodsecurity of households in the studysample.

The Trussell equations of the Brassmethod with Coale-Demeny WestModel life tables were used toestimate child survivorship andbackward extrapolation to estimatechild mortality rates at dates in therecent past.


Children aged 0to 5 yearsN= 4816households

Infant and childmortality

Receipt of food aid

Determinants of childsurvival

Infant and child mortality rateswere 941000 and 134/1000respectively.

HH-level demographic factors, HH-level food and livelihood security,community-level economicproduction, and access to potablewater were predictive of childsurvival.

HH receipt of food aid had a smallbut significant positive associationwith child survival, although eitherthe underlying causal mechanismsof this association or the role ofconfounding factors cannot bedetermined

It is remarkablethat the mostextensivedrought in thecountry’smodern historypassed withouta measurableincrease inchild mortalityamong thegeneralpopulation.


49

7. Summary of published studies assessing the impact of conditional cash transfers or vouchers in emergency situations

Authorof study

Objective of Study Setting Study design,Statistics

Surveypopulation

Indicator /outcome

Impact Conclusion

Mascue-Taylor et al(2010)

To determine whether acash-for-workprogramme during theannual food insecurityperiod in Bangladeshimproved nutritionalstatus in poor ruralwomen and children.

[In August and September2007, widespread severeflooding across northernBangladesh intensifiedand prolonged the impactof the food insecurityperiod (between mid-September and mid-November) for chars (largeflat islands in the mainriver channels) dwellers].

Chardwellingareas innorthernBangladesh(Sep- Dec2007)

A panel study which involveda random sample ofhouseholds from over 50 000enrolled in a cash-for-workprogramme and similarcontrol households. The height, weight and MUACof a woman and child fromeach household weremeasured at baseline and atthe end of the study (meantime: 10 weeks).

Women reported 7-dayhousehold food expenditureand consumption on bothoccasions. Changes inparameters were comparedbetween the two groups.


Women andchildren aged< 5y fromlandlesshouseholdswho wereliving on charsN= 895householdsparticipating inthe programme(intervention)

N= 921 nonparticipatingHH (controlgroup)

Weight gain

MUAC gain

HAZ, WAZ, WHZ

Baseline infoAt baseline, nosignificantdifference existedbetween thegroups.

By the end of the study, the differencein mean MUAC between women in theintervention and control groups hadwidened by 2.29 mm and thedifference in mean weight, by 0.88 kg. Among children, the difference inmeans between the two groups hadalso widened in favour of theintervention group for : height (0.08cm; P < 0.05), weight (0.22 kg; P <0.001), MUAC (1.41 mm; P < 0.001)and z-scores for height-for-age (0.02;P < 0.001), weight-for-age (0.17; P <0.001), weight-for-height (0.23; P <0.001) and mid-upper armcircumference (0.12; P < 0.001).Intervention households spent moreon food and consumed more protein-rich food at the end of the study.

The cash-for-workprogramme ledto greaterhousehold foodexpenditureandconsumptionand women’sand children’snutritionalstatusimproved.


50

Non-emergency situations

1. Summary of published studies assessing the impact of conditional cash transfers or vouchers in non-emergencies

Authorof study

Objective ofStudy


Surveypopulation

Indicator/ outcome

Impact Conclusion

Morris et al(2004)

To assess the impactof conditional cashtransfer programmeon growth ofchildren ofbeneficiaryhouseholdscompared to childrenof householdsaccidentallyprevented fromreceiving the benefit.

(In Brazil, a nationalhealth-relatedconditional cashtransfer programme,Bolsa Alimentac¸a˜o,aimed to reducenutritionaldeficiencies and toensure that high-riskhouseholds wereeffectively linked intothe National HealthService).

Four municipalities innortheast Brazil(Oct 2001 – Apr 2002)

- Observational study (crosssectional survey at the end ofprogramme ie after 6 mo. Aretrospective cohort analysisused a regression model todetermine whether children’sgrowth velocity accelerated atthe time their families began toreceive programme benefits.

Programme beneficiaries werecompared with matchedindividuals from householdsthat were originally selected toreceive the benefit but whosubsequently were excludeddue to 1 of 3 quasi-random,administrative errors

Monthly transfers ranged from15 to 45 Brazilian reais (US$6.25to US$18.70) per household permonth, depending on thenumber of eligible individualsfor 6 months.

All children <36 mo wereincluded forassessingimpact(programmeincludedchildren <7y)N= 630 ofwhichbeneficiaries=472 and nonbeneficiaries=158

Weight gain

Baseline infoNon provided

- Anthropometric status wasassessed 6 mo after benefitsbegan to be distributed, andbeneficiary children were 0.13Z-scores lighter (weight-for-age) than excluded children,after adjusting forconfounders (P = 0.024).

- The children’s growthtrajectories werereconstructed by copying upto 10 recorded weights fromtheir Ministry of Healthgrowth monitoring cards andby relating each weight to thechild’s age, gender, andduration of receipt of theprogramme benefit in arandom effects regressionmodel. Each additional monthof exposure to the programmewas associated with a rate ofweight gain 31 g lower thanthat observed in excludedchildren of the same age (P < 0.001).

This failure to respondpositively to theprogramme may havebeen due to aperception thatbenefits would bediscontinued if thechild started to growwell. Nutritionprograms should guardagainst giving theimpression that poorgrowth will berewarded.

Barber andGertler(2008)

To evaluate theimpact ofOportunidades, alarge-scale,conditional cashtransfer programmein Mexico, onbirthweight.

The programmeprovides cashtransfers to low-income, ruralhouseholds inMexico, conditionalon acceptingnutritionalsupplements healtheducation, andhealth care.

Seven Mexican states(2003)

Retrospective reports onrandomly assignedcommunities to treatment orcontrol group.

Intervention - Monthly healthstipend was fixed atapproximately US$ 15 perhousehold per month

Multivariate and instrumentalvariable analyses was used toestimate the impact of theprogramme on birthweight ingrams and low birthweight(<2500 g), receipt of any pre-natal care, and number ofpre-natal visits.

- Control group


Pregnantwomen (15-49y)

N= 506communities ofwhich 306made uptreatmentgroup.

Impact onbirthweightin grams

Theprobability oflowbirthweight.

Baseline infoprovided

Oportunidades beneficiarystatus was associated with127.3g higher birthweightamong participating womenand a 4.6 percentage pointreduction in lowbirthweight.

No p values or CI given

The Oportunidadesconditional cashtransfer programmeimproved birthweightoutcomes. This findingis relevant to countriesimplementingconditional cashtransfer programmes.

Lagard et al(2009)

To assess theeffectiveness ofconditional cashtransfers (CCT) inimproving access tocare and healthoutcomes, inparticular for poorerpopulations in lowand middle incomecountries.

Search strategy- A wide range ofinternationaldatabases, includingthe Cochrane CentralRegister of ControlledTrials (CENTRAL),MEDLINE andEMBASE, in additionto developmentstudies and economicdatabases weresearched. Websitesand online resourcesof numerousinternationalagencies,organisations anduniversities weresearched to findrelevant greyliterature. (Theoriginal searches Nov2005 - Apr 2006. Anupdated search inMEDLINE was carriedout in May 2009).

A qualitative analysis of theevidence was performed.

Selection criteriaCCT were defined as monetarytransfers made to householdson the condition that theycomply with some pre-determined requirements inrelation to health care. Studieshad to include an objectivemeasure of at least one of thefollowing outcomes: health careutilisation, health expenditure,health outcomes or equityoutcomes. Eligible studydesigns were: randomisedcontrolled trial, interruptedtime series analysis, orcontrolled before-after study ofthe impact of health financingpolicies following criteria usedby the Cochrane EffectivePractice and Organisation ofCare Group.

Mostly childrenbut alsopregnant andlactatingwomen

- Papersreportingresults from sixinterventionstudies.

N= 10 papers

Overall, designquality andanalysis limitedthe risks ofbias.

- Changes inuse of healthservices

- Changes inhealthoutcomes.

Several CCT programmesprovided strong evidence ofa positive impact on the useof health services,nutritional status and healthoutcomes, respectivelyassessed by anthropometricmeasurements and self-reported episodes of illness.

- It is hard to attribute thesepositive effects to the cashincentives specificallybecause other componentsmay also contribute. Severalstudies provide evidence ofpositive impacts on theuptake of preventive servicesby children and pregnantwomen. We found noevidence about effects onhealth care expenditure.

Conditional cashtransfer programmeshave been the subject ofsome well-designedevaluations, whichstrongly suggest thatthey could be aneffective approach toimproving access topreventive services.Their replicability underdifferent conditions -particularly in moredeprived settings - isstill unclear becausethey depend on effectiveprimary health care andmechanisms to disbursepayments.

Further rigorousevaluative research isneeded, particularlywhere CCTs are beingintroduced in lowincome countries, forexample in Sub-SaharanAfrica or South Asia


51

2. Summary of published studies of nutrition-related community-level interventions in non-emergency situations

Authorof study

Objective ofStudy


Surveypopulation

Indicator /outcome

Impact Conclusion

Colecraft etal (2004)

To examine theadoption of feedingrecommendationsamong caregiversof childrenrecuperating frommalnutrition andassess thedeterminants ofgrowth of childrenattending anutritionrehabilitationcentre (NRC) inAccra, Ghana.

NRCs at fourpolyclinics andparticipants’homes in Accra,Ghana.(Nov 1999- Jul2000).

- Longitudinal study inwhich attendance andmaternal programmeparticipation wererecorded daily andchildren’santhropometry anddietary intake weremeasured at four timepoints (admission,interim, exit, post-exit)at the NRC andparticipants’ homes.

- Post exit/follow upwas for up to 4 monthsafter leaving theprogramme.

No control group


Caregivers andtheir childrenreferred to anNRCN= 108caregivers and116 children

Attendance

Weight gain

Baseline infoChildren’s age was 12.7± 0.5 monthsThere were no significantcentre differences inchildren’s age, gender orbreast-feeding status.The majority of childrenwere underweight. Therewere no significantcentre differences in therates of wasting andunderweight; however,children referred to NRC1 and 4 weresignificantly more likelyto be stunted than thosereferred to NRC 2 and 3(P = 0.004).

Most caregivers attendedthe NRC sporadically(effective length of stay was1.4 ± 0.1 months). Use ofNRC-promoted foods in thehome after discharge waslow due to inaccessibility ofthe food items, lack ofpreparation knowledge ormoney, child preferencesand the common practice ofpurchasing ready-to-eatfoods.

Although there weresignificant increases inchildren’s weight-for-age (P= 0.048) and weight-for-height (P = 0.002) Z-scoresbetween enrolment anddischarge, most childrendiscontinued programmeparticipation beforeadequate recuperation.

The NRC education did notaddress the use of streetfoods for child feeding andwas unsuccessful inchanging in-home feedingbehaviour.

The prominence of streetfoods in children’s dietswarrants re-evaluation ofthe NRC’s educationalapproaches to enhancetheir responsiveness tocaregivers’ needs andeffectiveness for thecontinued recuperation ofmalnourished children athome. NRC feedingstrategies needimprovement to ensureadequate provision ofenergy and nutrients tosupport catch-up growthin children.

Bushamukaet al (2005)

To assess theadditional benefitsof a homesteadgardeningprogrammedesigned tocontrol vitamin Adeficiency inBangladesh

Hellen KellerInternational(HKI)initiated aprogramme called,“NGO Gardeningand NutritionEducationSurveillanceProject”(NGNESP),in 1993which aimed atencouraging poorhouseholds withvery limited land tostart gardening,producingvegetables year-round, andincreasing thenumber andproduction ofvitamin A–richcrops grown pergarden.

Rural Bangladesh(Feb-Mar 2002)

Observational study

According to the levelof householdparticipation in theprogramme, threegroups;- Active-participant(receiving technicaland material assistancefrom HKI for < 3y),- Former-participant(completed theprogramme andoperating without HKIassistance for at least3y), and- Control HH—wereincluded in the study

Data was collected onfood security and socialstatus of women.

- Control group


- Household foodsecurity wasdetermined byassessing theavailability of food tohouseholds, the abilityof households to accessfood, and theutilization of food byhouseholds.

- The social impact ofthe programme wasdetermined by thechanges, as perceivedby women, in theirability to contribute tohousehold livelihoodsand participate inhousehold decision-making.

Poor ruralwomenN= 2160 HHwith approx720 in eachgroup.

- Homestead gardenproduction used toassess the effect of theproject on foodavailability.

- Ability of householdsto access food -measured by theconsumption levels ofgarden produce

- Social impact of theprogramme

In a three-month period,the households of activeparticipants produced amedian of 135 kg andconsumed a median of 85kg of vegetables, while thecontrol householdsproduced a median of 46kg and consumed amedian of 38 kg (p <0.001).

About 64% of the active-participant householdsgenerated a mediangarden income of 347 taka(US$1 = 51 taka), whichwas spent mainly on food,and 25% of the controlhouseholds generated 200taka in the same period (p< .001). The gardenproduction and incomelevels of formerlyparticipating householdsthree years afterwithdrawal of programmesupport were much higherthan those of the controlhouseholds, illustratingthe sustainability of theprogramme and its abilityto increase household foodsecurity.

Significantly more womenin active- and former-participant householdsthan in control householdsperceived that they hadincreased their economiccontribution to theirhouseholds since the timethe programme waslaunched in their subdistricts (> 85% vs. 52%).Similar results were foundfor the level of influencegained by women onhousehold decision-making.

These results highlight themultiple benefits thathomestead gardeningprograms can bring anddemonstrate that thesebenefits should beconsidered when selectingnutritional anddevelopment approachestargeting poorhouseholds.

Cont’d next page


52

Authorof study

Objectiveof Study


Surveypopulation

Indicator /outcome

Impact Conclusion

Pappas et al(2008)

To evaluate theimpact of alarge-scaleintervention onthe nutritionalstatus andschoolenrolment ofprimary schoolgirls.

Twenty nine poorrural districts inPakistan(Apr 2003 – Oct2005)

The primarystrategy wasempowerment ofwomen in thecommunity whovolunteered toplan the meals,purchase the food,and cook and servethe meals.

Pre and postintervention study

Intervention providedfreshly prepared mealsin 4035 governmentprimary girls’ schoolsover a 2 year period.

Data was collected fromgrowth monitoring,attendance records, pre-and post-interventioncommunity based crosssectional surveys, focusgroup discussions, andthe use of otherethnographic methods.

A study on changes inthe levels ofmalnutrition was basedon an analytical sampleof girls who received atleast two sets of bodymeasurements at least6 months apart.

Primary schoolaged girls

N= 203 116

Stunting (HAZ < -2)Underweight (WAZ< -2)Wasting (WHZ < -2)

- School enrolment

- Dietaryknowledge

Baseline infoAt baseline, 23.2%girls were stuntedand 14.3% wasted.Reports of hungerfrom the schoolswere common.

Over the study period, improvementswere observed in the three measuresof nutritional status; wastingdecreased by 45% (p < 0.001),underweight girls declined by 21.7%(p < .001) and stunting decreased by6% (p < .001).

Dietary knowledge increased

School enrolment increased by 40%.Girls who entered the programmeearly were found to have similar levelsof malnutrition to girls who enteredlate, suggesting that factors externalto the programme were notassociated with the decrease inmalnutrition.

This studydemonstrates thepotential success andscalability of schoolfeeding programs inPakistan. Lessonslearned include thatsynergies are foundwhen working acrosssectors (health,education, andempowerment) andthat there arechallenges tointersectoral projects.

Globalization mayundermine thissuccessful model asPakistan considersexpanded schoolfeeding programs

Kidney et al(2009)

To provide asystematicreview of theeffectiveness ofcommunitylevelinterventionsto reducematernalmortality.

Worldwide Published papers weresearched usingMedline, Embase,Cochrane library,CINAHL, BNI, CABABSTRACTS, IBSS, Webof Science, LILACS andAfrican Index Medicusfrom inception or atleast 1982 to June2006; Unpublishedworks were searchedusing NationalResearch Registerwebsite, meta Registerand the WHOInternational TrialRegistry portal. Majorreferences were handsearched.

Maternity orchildbearingage women

- Community-levelinterventions

- Maternalmortality

Five cluster randomised controlledtrials (RCT) and eight cohort studiesof community-level interventionswere found. Results weresummarised as odds ratios (OR) andconfidence intervals (CI), combinedusing the Peto method for meta-analysis. Two high quality clusterRCTs, aimed at improving perinatalcare practices, showed a reduction inmaternal mortality reachingstatistical significance (OR 0.62, 95%CI 0.39 to 0.98). Three equivalenceRCTs of minimal goal oriented versususual antenatal care showed nodifference in maternal mortality(1.09, 95% CI: 0.53, 2.25). The cohortstudies were of low quality and didnot contribute further evidence.

Community-levelinterventions ofimproved perinatal carepractices can bringabout a reduction inmaternal mortality.This challenges theview that investment insuch interventions isnot worthwhile.Programmes toimprove maternalmortality should beevaluated usingrandomised controlledtechniques to generatefurther evidence.

Guyon et al(2009)

To improveinfant andyoung childfeedingpractices,increase uptakeofmicronutrientsupplements,and improvewomen’sdietarypracticesthroughimplementation of a nutritionproject basedon the EssentialNutritionActions (ENA)framework.

Six districts in twohighlandprovinces ofMadagascar(Feb 2000- Nov2005)

Observationaluncontrolledintervention study(Two cross sectionalsurveys fromrepresentativehouseholds wereconducted; one atbaseline and the otherat the end of theprogramme),

Interventions includedtraining, interpersonalcommunication,communitymobilization, and massmedia. Changes in practiceswere assessed througha comparison ofbaseline data andendline data.

- No control group


Children under2 years of ageN = 1,200 HH atbaseline and N = 1,760 HH atthe end ofprogrammeimplementation in 2005

- Rate of initiationof breastfeedingwithin 1 hour ofbirth.

- Rate of exclusivebreastfeeding ofinfants

- Iron-folic acidsupplementationduring pregnancyand postpartum VitA supplementation

- Maternal dietarypractises

The rate of initiation ofbreastfeeding within 1 hour of birthincreased from 32% to 68%, the rateof exclusive breastfeeding of infantsunder 6 months of age increasedfrom 42% to 70%, the rate ofcontinuation of breastfeeding at 20to 23 months increased from 43% to73%, the rate of feeding children theminimum recommended number ofmeals per day at 6 to 23 monthsincreased from 87% to 93%, the rateof iron–folic acid supplementationduring pregnancy increased from32% to 76%, and the rate ofpostpartum vitamin Asupplementation increased from17% to 54% (p <0.001 for allchanges).

Modest improvement was achievedin maternal dietary practices duringlactation and feeding of the sickchild after illness. The results wereinconclusive regarding food diversityfor complementary feeding.

No improvements were reported inincreasing food intake during childillness or pregnancy.

The ENA frameworkallows broad-scaleimprovement ofnutritional practices tobe achieved throughthe maximization ofcontacts using multipleprogrammeopportunities withinexisting health systemsand communitystructures and throughmass media.


53

Authorof study

Objectiveof Study


Indicator /outcome

Impact Conclusion

Agha (2004) To evaluate theeffectiveness ofasupplementaryfeedingprogramme innorthern Iraq

Nine primaryhealthcarecentres,Dohukprovince,northern Iraq(Jan - July2001)

Observational intervention studywith seven months follow up.Intervention – children receivedhigh-protein high-energy biscuitsin the first month and a monthlychild ration for preparingsoyabean mix throughout. Theirfamilies received food rations inthe first 4 months.(A smaller cohort ie some childrenwere evaluated right afterreceiving the high-protein high-energy biscuits for the firstmonth).

- No control group (but there wasa control for the smaller cohort)

- Inadequate statistics presented

Malnourishedchildren (childweight 2SD belowWHO standardweight-for-age)aged 6mo to 3y

N= 159 of which(27=interventiongroup and 27=normal group) forthe small cohort

Weight-for-age

Cohort Increase in weight

- Improvement was noticed for allchildren, particularly in the firstmonth.

- Problems with the rations andwithin the growth monitoring unitsresulted in significant drop- out. Useof standard growth charts may be away to overcome this problem.

CohortN= 24 in each group. The meandifference in weight increasebetween intervention and controlgroup was 152.1g, 95%CI: 16.4 to287.8, t-value = 2.241, p<0.05)

High energybiscuits shouldbe distributedthroughoutinstead of themix.

Hossain et al(2005)

To assess theeffect on foodintake ofaddingamylase-richflour (ARF)fromgerminatedwheat tosupplementary food amongchildren innine ruralCommunityNutritionCentres undertheBangladeshIntegratedNutritionProject (BINP).(Inadequatedietary intakeis a majorcontributor toprotein-energymalnutrition)

Nine BINPCommunityNutritionCentres,NarshingdiDistrict,Bangladesh(Dec 1998-Apr 1999)


- The centres were selectedrandomly from all CommunityNutrition Centres in the district,and were randomized in equalnumbers to three different diets: Diet 1The standard supplementaryfood (S-SF), Diet 2Supplementary food with addedwater (W-SF).Diet 3Supplementary food with addedARF (ARF-SF),(The composition of the diets wasthe same; however, theconsistency and calorie densitywere altered by adding either ARFor water).

- Intervention lasted for 6 weeksand diet was given 6 days perweek.

- No control group


Children aged 6 to24 mo who areseverelymalnourished(defined as W/A <60% of the NCHSmedian, orfaltering ingrowth i.e.,weight gainduring theprevious 2 months< 500 g forinfants and < 300g for children 13to 24 months).but withoutclinical evidenceof any acuteinfectious diseaseor morbidcondition thatcould interferewith the intake ofstudy diets,

N= 166 of whichS-SF group = 35, W-SF group = 66ARF-SF group= 65

- Supplementaryfood intake

- Vomiting

- Weight gain

- weight-for-length zscore increment

- Cost

Baseline infoThe mean ± SDweight-for-length zscores for S-SF, W-SFand ARF-SF groupswere -1.11 ± 1.03, -1.34 ± 0.78, 1.59 ±1.02 respectively

The mean ± SD intake ofsupplementary food from a singlemeal by children completing sixweeks on the diets was higher forchildren receiving ARF-SF (33.91 ±8.25 g) than for those receiving S-SF (25.66 ± 6.73 g) or W-SF (30.26± 8.39 g) (p < 0.05 for bothcomparisons).

The weight of vomited food wassignificantly higher for childrenreceiving W-SF than for children inthe other two groups (p<0.05).

Weight gain and increments inlength and weight-for-height werehigher for children who receivedARF-SF than for children in theother two groups, but thedifferences were not statisticallysignificant. The accept¬ability ofARF-SF was higher than that of thetwo other diets.

The additional cost of adding 2 g ofARF to the diet was about Taka0.25 (US$1 = Taka 48).

Addition of ARFto existingstandardsupplementaryfood, as usedunder the BINPprogramme, is asimple andeffective meansto increase theintake of food bychanging itsconsistency,thus making iteasier formalnourishedchildren toingest.

Kuusipalo etal (2006)

To examineacceptability,growth andchange inbloodhaemoglobinconcentrationamongmoderatelyunderweightambulatoryinfants givenfortifiedspread (FS).

Lungwena, arural area inMalawi (Nov 2002-Mar 2003)

A randomised, controlled,parallel- group, investigator-blind clinical trial.

Intervention: milk-based or soy-based fortified spread group into8 supplementation schemes;Nothing, 5, 25, 50 or 75 g/daymilk-based FS or 25, 50 or 75g/day soy-based FS.

Supplementation wasadministered at home weekly forthe first 4 weeks and fortnightlythereafter for the interventionperiod of 12 weeks.

- Control group


Underweightinfants aged 6 to17 months (W/A< -2), whoseweight is > 5.5kgand WHZ > -3.

N= 126

- Changes in weight,length and blood Hbconcentration.

Baseline infoInformation ispresented as mean (SD)for Hb (g/L),weight (kg),Length (cm)respectively forNothing 84(14), 7.7(0.8), 69.7 (3.2); 5gmilk-based FS 79 (15),7.3(1.0), 69.1 (3.4); 25gmilk-based FS 79 (12),7.3 (1.0), 69.0(3.9); 50gmilk-based FS 77 (10),7.3(1.3), 69.0 (4.6); 75gmilk-based FS 81(16),7.7 (0.4), 71.1 (1.5);25g soy-based FS77(20), 7.5 (0.9), 69.9(3.7); 50g soy-based FS73(11), 7.4(0.9),69.1(3.4); 75g soy-based FS 79(22), 7.6(0.4), 69.9 (1.7).

One hundred and twenty five infantscompleted the intervention. Allinfants accepted the spread well,and no intolerance was recorded.

Average weight and length gainswere higher among infants receivingdaily 25 to 75 g FS than among thosereceiving only 0 to 5 g FS.

Mean Hb conc remained unchangedamong unsupplemented controlsbut increased by 10 to 17 g/L amonginfants receiving any FS. All averagegains were largest among infantsreceiving 50 g of FS daily: meandifference (95% confidence interval)in the 12-week gain between infantsin 50g milk-based FS group and theunsupplemented group was 290 g(range, -130 to 700 g), 0.9 cm(range, -0.3 to 2.2 cm), and 17 g/L(range, 0 to 34 g/L) for weight,length and blood Hb concn,respectively.

In soy- vs milk-based FS groups,average outcomes were comparable.

Supplementationwith 25 to 75g/day of highlyfortified spread isfeasible and maypromote growthand alleviateanaemia amongmoderatelymalnourishedinfants.

Further trialsshould test thishypothesis.

Cont’d next page

3. Summary of published studies assessing the impact of moderate acute malnutrition programmes in non-emergencies


54

Authorof study

Objective ofStudy


Surveypopulation

Indicator /outcome

Impact Conclusion

Gartner et al(2006)

To assess individualdeterminants ofdifferential benefitfrom the SenegalCommunityNutrition Project(CNP) by monitoringimprovement inchildren’s weight-for-age index orunderweight status(WA < –2 Z-scores)during participation.

(One of the objectivesof the CNP nutritioncomponents was tohalt furtherdeterioration inchildren’s nutritionalstatus in poor urbanneighbourhoods)

Diourbel, Senegal(dates notspecified)

Follow-up study usingthe CNP childmonitoring data. Linear general modelscompared variations inW/A according to 14factors describing thebeneficiaries and CNPservices.

Intervention – providedtargeted children withmonthly growthmonitoring andpromotion as well asweekly foodsupplementationprovided that mothersattended weeklynutrition educationsessions.

Intervention lasted for 6months.

- No control group


Underweightchildren (W/A <-2 z scores ofWHO/NCHSreferencemedian) ornutritionally at-risk (sibling ofan underweightchild or nothaving gainedweight duringthe last threemonths) aged6–35 mo whoparticipated inthe first twoyears of theproject

N= 4084.

- Recovery

- Characteristics ofparticipatingchildren and theirmothers whichcould influence theeffect of the CNPservices on themain processoutcome(improvement inanthropometricstatus ofparticipatingchildren)

Baseline infoTwenty eightpercent of children(6-35mo) in thearea wereunderweight

Mean W/A varied from 22.13 (SD0.82) to 21.58 (SD 0.81) Z-scoresbetween recruitment and the endof the follow-up. The lower thechild’s initial W/A, the greater wastheir increase in W/A but the lowerwas the probability of recoveryfrom underweight. Only 60.7%(95% CI: 58.6–62.9%) ofunderweight children recovered.

Six months of CNP services may notbe sufficient for catch-up growth ofseverely underweight children. Thenumber of food supplement rationsreceived was not a direct indicatorof the probability of recovery. Afteradjustment for services receivedand initial W/A, probability ofrecovery was lower in girls, inyounger children, in twins andwhen mothers belonged to aspecific ethnic group.

Determinants ofbenefit from CNPdiffered from the riskfactors forunderweight.Identification ofparticipants with alower probability ofrecovery can helpimprove outcome.Moreover, anexplanation for thelack of recovery couldbe that manyunderweight childrenare stunted but notnecessarily wasted.

Nahar et al(2008)

To i) assess whetherthe BangladeshIntegratedNutritionProgramme (BINP)correctly identifiedwhich pregnantwomen should beenrolled in the foodsupplementationprogramme, ifsupplementationcommenced ontime and was takenon a regular basis.

ii) determinewhether foodsupplementationled to enhancedpregnancy weightgain and reductionin the prevalence oflow birth weight.

A rural union ofBhaluka Upazila,Mymensingh,Bangladesh (From1995, end datenot stated)

- A one-yearcommunity-basedlongitudinal study.

- The food supplementprovides 2512 kJ (600kcal)/d

- Supplement foodwas prepared at thevillage level. Pregnantmothers had toconsume thesupplement at a localCNC.

- Control group (nonsupplementedpregnant women)


Normotensive,non-smokingpregnantwomen whoregisteredbetween 2ndand 6th monthof gestation. N= 1104 ofwhichsupplementedwomen = 358and nonsupplementedwomen = 746

BINP programmaticissues(i) receivingsupplementation ifBMI<18.5 kg/m2(ii) commencingsupplementation atthe correct time(iii) receiving dailysupplementationsix days a weekuntil the birth ofthe child (iv) achievingtargeted weightgain set by BINP of>1 kg per month or>7 kg at the end ofpregnancy.

Pregnant women who had a BMIof < 18.5 kg/m2 on firstpresentation should have beenselected for supplementaryfeeding (2512 kJ (600 kcal)/d forsix days per week) starting atmonth 4 (16 weeks) of pregnancy.However, of the 526 women whohad BMI < 18.5 kg/m2, only 335received supplementation; so thefailure rate was 36.3%. Inaddition, of those receivingsupplementation, only 193women (36.7% of 526 women)commenced supplementation atthe correct time, of whom thirty-two (9.6% of 335 women)received supplementation for thecorrect number of days (100%days).

There were no significantdifferences in mean weight gainbetween BMI <18.5 kg/m2supplemented or non-supplemented groups or betweenthe equivalent groups with BMI ≥18.5 kg/m2. Weight gain wasinversely related to initial weight,so lighter women gainedrelatively more weight duringtheir pregnancy than heavierwomen. The mean birth weight inthe supplemented and non-supplemented groups was 2.63 kgand 2.72 kg, respectively. Motherswith BMI < 18.5 kg/m2 who wereor were not supplemented hadalmost equal percentages of low-birth-weight babies (21% and 22%, respectively).

The study raisesdoubt about theefficiency of the BINPto correctly targetfoodsupplementation topregnant women. Italso shows that foodsupplementationdoes not lead toenhanced pregnancyweight gain nor doesit provide anyevidence of areduction inprevalence of lowbirth weight.

Cont’d next page


55

Authorof study

Objective ofStudy


Indicator /outcome

Impact Conclusion

Augusto etal (2010)

To evaluate theeffectiveness of agovernmentalsupplementaryfeeding programme(fortified milkdistributionprogramme) onchild weight gain.

São Paulo, Southeastern Brazil (Jan2003 to Sep 2008).

A cohort study including secondarydata on 25,433 low-income children.Children were divided into three z-score groups when starting theprogramme:Group 1Weight gain not compromised (WNC)(z > -1) Group 2Risk of low weight (RLW) (-2 ≤ z< -1)Group 3Low weight (LW) (z <-2).

Intervention - 15 litres of milk permonth per participating child,equivalent to 500 mL/day.

- Children were measured every 4mo until the end of the programme

- No control group


Children aged 6to 24 mo (withno healthproblems) fromfamilies with amonthlyincome lowerthan twominimumwages.

N= 25,433

- Weight gain,measured usingweight-for-age z-score values

Baseline info(weight-for-age z-score) whenregisteringWNC group =82.4%,RLW group= 14.3%

LW group = 3.3%

The programme had apositive effect on childweight gain, varyingaccording to childnutritional status whenstarting this programme;among those who startedit with weight gain notcompromised, the meanadjusted gain z-score was0.183 95% IC (0.175,0.191); among those withrisk of low weight, 0.56695% CI (0.546, 0.586);and among those withlow weight, 1.005 95%CI(0.954, 1.056).

The programmewas effective forweight gain inchildren youngerthan two years,with a morepronounced effecton children whostart theprogramme underless favourableweight conditions.

Flax et al(2010)

a) To compare childfood consumptionpatterns andcaregiver and childfeeding behavioursin two study groupsreceiving eitherlipid-based nutrientsupplements (LNS)or corn-soy blend(CSB).

b) To examinebehaviours relatedto the use of plainLNS versus LNSmixed withporridge.

(the way caregiversuse supplementaryfood forundernourishedchildren andintegrate it intofeeding patternsmay influence thebenefits achieved bysupplementation)

Mangochi district,southern Malawi (April to July 2005)

Intervention trial in which childrenwere randomised into 2 groups

Intervention -supplementationwith either LNS or CSB for 12weeks.

- Observational data werecollectedduring one 11h home visit perparticipant.

- Semi-structured forms were usedto collect information about childand caregiver behaviour duringsupplement feeding episodes.

- No control group


Underweightchildren aged6–17 mo (WAZ< -2)N= 170 ofwhichLNS group = 85and CSB = 85

Baseline infoMean (SD) WAZwas -2.9 (0.7) inthe LNS groupand -3.1 (0.8) inthe CSB group.

- Number offeeding episodeswhich include: startand end times offeeding and type offood given.

- Caregivers’behavioursmeasured included:hand washingbefore the feedingepisode, personfeeding the child,utensil used,whethersupplement wasshared, whetherthere was anyleftover at the endof the feedingepisode, and theproportion sharedand leftover.

- Weight gain

There was no observeddifference between theCSB and LNS groups inmean number of feedingepisodes per day ormean daily feedingtime.

- Caregivers in the CSBgroup were significantlymore likely than those inthe LNS group to servethe supplement as meal,wash their hands beforefeeding, and feed thechild with a spoon.

- The mean proportionof the offeredsupplement that waseaten by participantswas slightly higher inthe LNS group (CSB 94%,LNS 98%, difference(95% CI) 4% (1%, 7%), p= 0.003).

There were significantlymore instances ofleftovers and the meanproportion leftover washigher in the CSB thanthe LNS group (CSB 19%,LNS 6%, difference (95%CI) 12% (9%, 15%), p <0.001).

- Sub-groupcomparisons ofbehaviours related tothe use of LNS mixedand plain generallymirrored the results ofCSB compared to LNS

- Presence of leftoverswas negativelyassociated with changein child’s WAZ. The otherexplanatory variablesentered to the model(number of feedings,caregiver hand washing,and supplementsharing) did not predictchange in child’s WAZ.

Programspromoting LNS inMalawi shouldconsider behavioursrelated to mode ofserving and provideadvice to caregiversin order tominimize leftoversduring supplementuse.

Cont’d next page


56

Authorof study

Objective ofStudy


Indicator /outcome

Impact Conclusion

Singh et al(2010)

To evaluate theeffectiveness of alocally madeready-to-usetherapeutic food(RUTF) indecreasing mildto moderatemalnutrition.

Vellore, India(Jan to May,2008)

A randomized open label,controlled trial.Intervention - A locally producedenergy-dense supplement (RUTF)administered in pre-schools byteachers, compared to the currentstandard of care [teachingcaregivers how to make a fortifiedcereal-milk supplement calledHigh Calorie Cereal Milk (HCCM)].

3 months follow-up from date ofrecruitment. Measurements weretaken on day 1, day 30, 60 and 90,with a window of 5 days after thescheduled date

- No control group


Children aged 18-60 months withW/A ≤ 2 SD. N= 118 of whichRUTF= 61 andHCCM= 57

- Increase in WAZ> -2

- Increase in levelsof plasma zinc,vitamin B12, serumalbumin andhaemoglobin.

Baseline infoMean (SD) weight(kg)= 10.79 (1.28)for RUTF group and10.98 (1.34) forHCCM group

The Mean (SD) weight gain at 3months was higher in the RUTFgroup: RUTF (n=51): 0.54 kg; (SE= 0.05; 95% CI = 0.44 – 0.65) vsHCCM (n=45): 0.38 kg; (SE =0.06; 95% CI = 0.25 – 0.51), P =0.047.

- The weight gain per kilogramof body weight was directlyproportional to the severity ofmalnutrition.

Significant increases in serumalbumin was seen in both theRUTF [P=0.045] and the HCCM[P=0.027] groups, as werechanges related to anaemia, withimprovement in both groups.

- The cost of one month’s supplyof RUTF, at the rate of 250 g perweek, or 50 g per week day, foreach child was calculated to beapproximately INR 135 (USD2.95).

Community-basedtreatment showedweight gain in bothgroups, the gainbeing higher withRUTF.The cost-effectiveness of thismethod of therapyneeds to beevaluated.

Jillcot et al(2010)

To describe thedevelopment,implementation,evaluation, andsubsequentimprovements ofa supplementalfeedingprogramme thatprovidescommunity-based care tounderweightchildren in a ruralEast Africansetting, using alocally-sourcedand producedready-to-usefood (RUF)

Bundibugyo,Uganda(Oct-Nov2007)

Observational interventionstudy.

Intervention - ground soybeansand groundnuts (peanuts) mixedwith moringa oleifera leafpowder to form an energy-densesupplemental food, designed foruse as an RUF for 5 weeks.

Programme was evaluated byexamining RUF nutrientcomposition, weight gainvelocity, and qualitative datafrom key-informant interviewsand home feeding observations.

- No control group


Children aged 6to 59mo withW/A < the 3rdpercentile line onthe standardUgandanimmunizationcard (whichincludes a growthchart) and/orMUAC < 12 cm.

N=Approximately 20

- RUF NutrientComposition

- Weight gain

- Feedingobservations

Baseline infoProvided in atabular form buttoo cumbersome tobe presented here

- Locally-produced RUF hadsimilar energy density 5.3kcal/g (21.5 kJ/g) but higherprotein content thancommercial RUTF (22 kJ/g).

- Mean weight gain of childrenwas 2.5 g/kg/day (range 0.9–6.0).

- Feeding observationsrevealed that caregivers werediluting the RUF fed to children.Production team membersdesired increased financialcompensation for their work butwere enthusiastic about theprogramme as helpful tomalnourished children.

Locally-producedRUF is a promisingstrategy forcommunity-basedcare of moderatelymalnourishedchildren. Throughthe productionteam’sentrepreneurship, asmall business wasformed, wherebyfinancial incentivesencouragedcontinued RUFproduction. Futureefforts are neededto educatecaregivers on correctRUF use andimprove commercialviability in localmarkets.

Lagrone et al(2010)

To determine theoperationaleffectiveness oftreatingmoderate acutemalnutritionwith ready-to-usesupplementaryfood.

(A randomisedclinicaleffectiveness trialin 2007 showedcorn-soy blend tobe less effectivethan RUSF)

RuralsouthernMalawi

Observational interventionstudy

Each child received 65 kcal/kg/dof locally produced soy/peanutRUSF, a product that providedabout 1 RDA of eachmicronutrient.

- Anthropometricmeasurements were taken every2 weeks and additional rationsof RUSF were distributed at thistime if the child remainedwasted.

- Study participation lasted upto 8 weeks.

- No control group


Children aged 6-59 mo with MAM(WHZ between -2& -3 withoutbilateral pedaloedema)N= 2417

Recovery

DefaultedFailedDied

Cost

Baseline infoMean (SD) of WHZwas -2.4 (0.6).

Eighty percent recovered, 4%defaulted, 0.4% died, 12%remained moderately wastedand 3% developed SAM.

Weight, length and MUAC gainwere 2.6g/kg/d, 0.2mm/d and0.1mm/d respectively.

Cost per child treated was $5.39

This interventionproved to be robust,maintaining highrecovery rates andlow default rateswhen institutedwithout theadditionalsupervision andbeneficiaryincentives of aresearch setting.


57

Authorof study

Objective ofStudy


Indicator /outcome

Impact Conclusion

Amadi et al(2005)

To assess theeffect of anexclusive diet ofamino acid-basedelemental feed(AAF) to standardnutritionalrehabilitation forthe treatment ofpersistentdiarrhoea-malnutritionsyndrome (PDM).

UniversityTeachingHospital inLusaka,Zambia(1998-2000)

- Randomised controlled trial.Treatment was given for 4 weeksin a single-blind study.

- Statistical tests presented

Malnourishedchildren (definedusing the Wellcomeclassification) withpersistent diarrhoea(defined as durationof 14 days or more)aged 6 to 24 mo.N= 200

- Weight gain

- Recovery fromdiarrhoea

- Mortality

Baseline infoAt randomization,median baselineWAZ was -4.0(interquartilerange, IQR -4.4, -3.5); 9% wereunderweight, 23%had marasmus,47% kwashiorkorand 21% hadmarasmic-kwashiorkor.

Weight gain was greater in theAAF group (median gain inWAZ was 1.23, IQR 0.89-1.57)compared with the controlgroup (0.87, IQR 0.47-1.25;P=0.002), although calorieintakes were higher in thecontrol group.

The increase in Hb conc. wasalso greater in the AAF group(0.8 g/dl, IQR 0-1.8) than in thecontrol group (0.3, IQR -0.6,-1.6; p=0.04).

Diarrhoea frequency andglobal recovery scoresimproved equally in bothtreatment groups andmortality did not differ.

A diet of reducedmolecularcomplexity wasassociated withsignificantlyimproved weightgain.

Elementalfeeds should not beadopted fortreatment of thesechildren due to theirexpense; rather it isnow important toestablish whetherlocally producedfeeds of reducedmolecularcomplexity canachieve the samebeneficial effect as asynthetic elementalfeed.

Ciliberto etal (2006)

To assess theeffect ofexclusive home-based therapy onchildren withoedematousmalnutrition.

Malawi (Dec 2002-June 2004 orNov 2003 -Apr 2004)

- Prospective uncontrolled caseseries of 2 therapeutic nutritionalstudies. (Participants were drawnfrom 2 large previously reportedcommunity-based malnutritionstudies, one in whichoedematous malnutrition caseswere detected passively andanother actively).

Intervention was RUTF andparticipation lasted 8 weeks.Length was measured every 4weeks

- No control group


Oedematousmalnourishedchildren aged 12 to60 months.

N=219 of which 171were from passivedetection and 48from activedetection

Wasting was definedas WHZ </-2.

- All children withmild oedema(defined as < 0.5 cmof pitting oedemaon the dorsum of thefoot) and a goodappetite weretreated exclusivelyas outpatients andincluded.

- Recoverydefined asresolution of theoedema for 4 wksand attainment ofWHZ > -2

- Failure to gainweight or relapse

- Death

Baseline infoNot given

- The overall recovery ratewas 83% (182/219), and thecase-fatality rate was 5%(11/219).

- For children with wastingand oedematousmalnutrition, 65% (55/85)recovered and 7% (6/85) died.

- The average weight gainwas 2.8 ±3.2 g/kg/d (mean±SD).

- Enrolment WHZ wasidentified as a significant riskfactor for death usingregression modelling(standardized coefficientβ=0.16, t stat =2.85,p<0.01).

- The estimated number ofdeaths using the method ofPrudhon was 7 (95%CI: 3-11).

This preliminaryobservationsuggests thatchildren withoedematousmalnutrition andgood appetite maybe successfullytreated with home-based therapy; aRCT to evaluate thisis warranted.

McLennanand Mills(2006)

To assess theimpact of anexisting partialdayrealimentationprogramme onthe growth ofchildren withprotein energymalnutrition(PEM). It washypothesizedthat at least 50%of the childrenwould achieve aminimumrecommendedgrowth rate of 5g/kg/day andwould attend atleast 50% of thepossibletreatment days.

A day hospitalin SantoDomingo,DominicanRepublic. Thisprogrammeoperates Monthrough Fri, 8a.m. to 12:30p.m. (2 Jul 2004 -30 Nov,2005).

- Observational interventionstudy.

- Caregivers were interviewedand child anthropometrics wereobtained at baseline and follow-up.

Clinical attendance patterns wereextracted from medical records.

Intervention – treatment was asper WHO guidelines. Childrenconsumed 2–3 feeds at thecentre and an additional 2–3feeds were sent home with theguardian for afternoon and earlyevening feeding. No supplieswere distributed for weekends.

Intervention lasted until childattained ≥ -1 SD of median W/H

- A follow-up feedbackquestionnaire was completed at12 weeks post admission.

- No control group


All childrenconsecutivelyadmitted to thepartial daytreatmentprogramme withsevere or moderatewasting(Age not specified)N= 88

- Weight gain

- Attendance

Baseline infoMean (SD) childage (months) was29.6 (29.8).W/H ≤ -1Z, ≥ -2Zand W/H ≤ -2Zwas 10.2 (9) and78.4 (69)respectively

Mean rate of weight gain inthe rehabilitation phase up to4 weeks following admissionwas 3.9 (SD 4.5) g/kg/daywith only 27% of the childrenachieving a minimumrecommended rate of > or =5.0 g/kg/day. On consecutiveclinic attendance days, themean growth rate was 4.2(SD8.6) g/kg/day, while on non-attendance days it wasapproximately 3.7 (SD 4.5)g/kg/day.

Children attended 80% of thepossible clinic days during thefirst 4 weeks of treatment.Within this time, 20% achievedthe target of > or = -1 SD ofthe median weight for height.Caregivers reported havingdifficulty finding caretakers fortheir other children and theirown illnesses as barriers toregular attendance.

There was asubstantialvariation in growthrates of childrenattending the clinicwith mean growthrates failing toachieve minimalstandards. Thoughsome children mayhave benefited fromthe partial daytreatmentprogramme,alternativestrategies should beconsidered at thisclinic to improveresource utilizationand outcomesincluding the use ofa home recoveryoption and anenhanced daytreatmentprogramme.

Cont’d next page

4. Summary of published studies assessing the impact of severe acute malnutrition programmes in non-emergency situations


58

Authorof study

Objective ofStudy


Surveypopulation

Indicator /outcome

Impact Conclusion

Ashworth(2006)

To examine theeffectiveness ofrehabilitatingseverelymalnourishedchildren in thecommunity innonemergencysituations.

The context isa routinehealthsystem withprimaryhealth-careprovision andreferralopportunities,in anonemergencysetting.

A literature search wasconducted (using acombination of databasesearches and hand-searching) of community-based rehabilitationprograms delivered by day-care nutrition centers,residential nutritioncenters, primary healthclinics, and domiciliary carewith or without provisionof food, for the period1980–2005.

Severelymalnourishedchildren (WHZ< -3 SD usingWHO definitionor the presenceof oedema).Age notspecified

N= 33 studiesof community-basedrehabilitation

- Effectiveness(defined as mortalityof less than 5% and anaverage weight gainof at least 5 g/kg/day).

Eleven (33%) programs wereconsidered effective.

Of the subsample of programsreported since 1995, 8 of 13(62%) were effective. None ofthe programs operatingwithin routine health systemswithout external assistancewas effective.

ConclusionWith careful planning andresources, all four deliverysystems can be effective. It isunlikely that a single deliverysystem would suit allsituations worldwide.The choice of a systemdepends on local factors.High energy intakes (> 150kcal/kg/day), high proteinintakes (4–6 g/kg/day), andprovision of micronutrientsare essential for success.

When done well,rehabilitation at home withfamily foods is more cost-effective than inpatient care,but the cost effectiveness ofready-to-use therapeuticfoods (RUTF) versus familyfoods has not been studied.

Where children have access toa functioning primary health-care system and can bemonitored, the rehabilitationphase of treatment of severemalnutrition should take placein the community rather thanin the hospital but only ifcaregivers can make energy-and protein-dense foodmixtures or are given RUTF.

For routine health services,the cost of RUTF, logistics ofprocurement and distribution,and sustainability need to becarefully considered.

Ashraf et al(2007)

To evaluate aday-care clinicapproach thatprovidedantibiotics,micronutrientsand feedingduring the daywith continuedcare by parentsat home at nightin treating severemalnutrition asan alternative tohospitalization.

(Some childrenare unable to beadmitted to ahospital despiteappropriatereferral, forvarious reasonsincluding lack ofhospital beds,inability of theparents to visit ahospital due tolong distance orfinancial or otherdomestic reasonssuch as the needto care for siblingsat home, and theneed for themother to work).

Dhakar,Bangladesh(Feb 2001 -Nov 2003)

- A prospectiveobservational study

- Participants deniedadmission to hospitalwere enrolled at the day-care nutritionrehabilitation (NR) unit ofRadda Clinic and receivedprotocolizedmanagement withantibiotics,micronutrients and milk-based diet from 8:00 amto 5:00 pm daily whilemothers were educatedon continuation of care athome. This continueduntil children attained80% weight-for-length.

- Follow up was eachweek for 2 weeks, thenevery 2 weeks up to 3months; the last 50% ofenrolled children wasmonthly for 6 mo.

*The management at theRadda Clinic was carried out inthree phases: acute phase, NRphase and follow-up phase

- No control group


Severelymalnourishedchildren(weight-for-length ≤ -3 SDand/or, bi-pedalnutritionaloedema or W/A≤ -3 SD withacute illness)aged 6–23 mo

N= 264

Duration of stay

Weight gain

Baseline infoMean (SD) of weight-for-length z score andWAZ was -2.7 (0.9)and -4.5 (0.8)respectively. Also52% were boys and78%, 21% and 1% hadmarasmus,marasmus-kwashiorkor andkwashiorkor,respectively. Only 13%had severemalnutrition alonewhile 35% hadpneumonia, 35% haddiarrhoea and 17%had both pneumoniaand diarrhoea.

At discharge, mean (SD) ofweight-for-length z scoreand WAZ was -1.7 (0.5) and -3.9 (0.7) respectively.

Mean (SD) duration of acuteand NR phases were 8 (4)and 14 (13) days,respectively.

Children gained weight[mean (SD) g/kg day] morerapidly during acute 10 (7)than NR phase 6 (5).Successful management waspossible in 82% (95% CI: 77–86%) children, 12%discontinued treatment and6% referred to hospitals.Only one child died duringNR phase.

Severely malnourishedchildren can be successfullymanaged at existing day-care clinics using aprotocolized approach.

Hossain et al(2009)

To compare theeffectiveness oflocally adaptedInstitute of Childand MotherHealth (ICMH)protocol with theWHO protocol forthe managementof severelymalnourishedchildren inBangladesh.

Two hospitalsin Dhaka,Bangladesh(Jun-Dec2003)

- Quasi-experimentalnon-randomized clinicaltrial.

Intervention - childrentreated with either WHOprotocol (Group I) or ICMHprotocol (Group II).

- No control group


Severelymalnourishedchildren (2-59mo) withweight forheight <70% ofWHO/NCHSreference withor withoutbilateral pittingoedema

N= 60 of whichgroup I =30 andgroup II = 30.

- Clinicalimprovement

- Weight gain

- Time taken toachieve target weightgain and

- Mortality

Baseline infoMarasmus, marasmus-kwashiorkor andkwashiorkor were66.8%, 13.3% and20% in group I and66.8%, 16.7%, 16.7%in group IIrespectively.

Mean (SD) weight related togain in Group I and Group IIwas 11.2 (4.1) and 11.1 (3.9)g/kg/day, respectively. Theweight gain was not relatedto the age group or type ofmalnutrition. The time takenfor oedema to subside (7.3 dvs 8 d) and for improvementof appetite (6.5 d to 7.3 d vs6.7 d to 8.4 d) was similarbetween the groups.

The target weight gain wasachieved in 28.3 (11.5) daysin Group I against 27.9 (6.2)days in Group II (P=0.88).The mortality rate was 6.7%in each group.

Treatment of severemalnutrition with locallyadapted ICMH protocol usinglocally available foods is asefficacious as the WHOprotocol.

Cont’d next page


59

Authorof study

Objectiveof Study


Surveypopulation


Bachman(2009)

To (i) describe theoutcomes ofCTC in Lusaka (ii) estimate thecost of CTC and(iii) estimatetheeffectivenessand costeffectiveness ofCTC for treatingchildren withSAM comparedwith notreatment.

Primaryhealth carecentres inLusaka,Zambia (2005-2008)

- Cohort study

Cost effectiveness wasbased on a decision treemodel considered fromthe perspective of healthservices.

- The existing model ofcare was compared to ahypothetical alternativeof providing notreatment (control group)

- Statistics presented.

- Severelymalnourishedchildren <5y(defined asMUAC ≤ 11cmor bilateralpittingoedema).

N= 21 LusakaDistrict HealthManagementTeam.

- Costs

Probabilities of outcomes

- Mortality up to year afterdeveloping SAM

- Expected DALYs after survivingone year.

The mean cost of CTC per childwas $203 (95%CI: $139-$274),of which RUTF cost 36%,health centre visits cost 13%,hospital admissions cost 17%and technical support whileestablishing the programmecost 34%.

Expected death rates within 1yof presentation were 9.2%with CTC and 20% with notreatment (risk difference11.5% (95%CI: 0.4 – 23.0).

CTC cost $1760 (95%CI: $592-$10142) per life saved and $53(95%CI: $18-$306) per DALYgained.

CTC was at least 80% likely tobe cost effective if society waswilling to pay at least $88 perDALY gained.

CTC is relatively costeffective comparedto other priorityhealth careinterventions indevelopingcountries, for a widerange ofassumptions.

Bahwere etal (2009)

To assessacceptabilityandeffectiveness ofa locally madeready-to-usetherapeuticfood (RUTF) inHIV-infectedchronically sickadults (CSA)with mid-upper-armcircumference< 210 mm orpittingoedema.

Home-basedcareorganizations(HBC) in thecentral regionof Malawi(May-Oct2005)

A prospective descriptivestudy

- Participants wereprescribed 500 grepresenting ~2600kcal/day of locally madeRUTF for three monthsand routinecotrimoxazole.

Weight, height, MUAC,Karnofsky score andmorbidity were measuredat admission and atmonthly intervals. Theamount of RUTF intakeand acceptability wereassessed monthly.

- No control group


Wasted AIDSadults (MUAC <210, BMI < 17kg/m2 bilateralpitting oedemaof the feet orlegs orbedridden for >50% of thedaytime formore than onemonth.N= 60

- Total RUTF intake

- Weight, MUAC, BMI gain

- Physical activity performance

Baseline infoThe median age of the patientswas 37.7 years (29.7–44) and63.3% of them were female.Clinically, 75% had WHO HIV stage4 and 25% HIV stage 3, but HIVinfection was confirmedserologically at the time ofinclusion in only 55% (33/60). Allpatients had reduced physicalactivity performance.

Overall average weight was 41.6(7.3) kg, 23.3% (14/60) of patientshad oedema, 78.3% (47/60) had aMUAC of less than 210 mm(overall mean ± SD of 195.0 ±18.8 mm), 69% (40/58) had a BMIof less than 17 kg/m2 (overallmean ± SD of 16.1 ± 1.7 kg/m2)and 50% (29/58) a BMI less than16 kg/m2.

- Mean daily intake was 300g/person/day (~1590 Kcal and40g of protein)

- Overall, 73.3% (44/60)gained weight, BMI, andMUAC. The median weight,MUAC and BMI gains afterthree months were 3.0 kg, 25.4mm, and 1.1 kg/m2,respectively.

- The total RUTF intakepredicted the MUAC, weightand BMI changes (Pearsoncorrelation coefficient = 0.503(p < 0.001), 0481 (p < 0.001)and 0.458 (p = 0.001),respectively).

The intervention improved thephysical activity performanceof participants and 78.3%(47/60) regained sufficientstrength to walk to the nearesthealth facility. Mortality atthree months was 18.3%(11/60).

Locally made RUTFwas acceptable topatients and wasassociated with arapid weight gainand physical activityperformance. Theintervention is likelyto be more costeffective thannutritional supportusing usual food-aidcommodities.

Fergussonand Tomkins(2009)

To explore HIVprevalence andmortality inchildren withSAM in sub-Saharan Africabysystematicallyreviewing therelevantpublishedliterature andpooling datafor meta-analysis

All studiesreporting onHIV infectionwithin asample ofchildren withSAM(Searcheswereconductedfrom Feb-Jun2008)

This review wasconducted withreference to the Meta-analysis OfObservational Studies inEpidemiology (MOOSE)group’srecommendations.

Electronic searches wereundertaken in MEDLINE,EMBASE, ScienceCitation Index (SCI) andthe Cochrane Library

Reference lists ofselected articles weresearched as well asrelevant conferenceabstracts, editorials,expert opinions andletters to journal editors.A draft of the article wascirculated among theauthors cited to checkfor data accuracy andcompleteness.

Children withSAM having HIVinfection (HIVstatus wasassessed using ablood test andSAM wasdefined usingthe WHO,Gomez,Wellcome orWaterlowdefinitions).

N= 4891children from17 peer-reviewedpapers from 15studies.

- HIV prevalence

- Mortality

- Inpatient care versus CTC

Of the 4891 children includedin the analysis, 29.2% wereHIV-infected.

HIV-infected children weresignificantly more likely todie than HIV-uninfectedchildren (30.4% vs. 8.4%; P <0.001; relative risk = 2.81,95% CI: 2.04 -3.87).

HIV negative children treatedwithin community-basedtherapeutic care (CTC)programmes had lowermortality (4.3%) than thosetreated within an inpatientnutrition rehabilitation unit(NRU) (15.1%), (p< 0.001).

There was no significantdifference in mortality forHIV-infected children withSAM treated in the CTC(30.0%) or NRU (31.3%)settings.

HIV prevalence ishigh in childrenwith SAM in sub-Saharan Africa, andHIV-infectedchildren are atsignificantlyincreased risk ofmortality.There is an urgentneed to integrateHIV testing andtreatment into carefor children withSAM in regions ofhigh HIVprevalence.

Cont’d next page


60

Authorof study

Objectiveof Study


Surveypopulation

Indicator /outcome

Impact Conclusion

Giugliani etal (2010)

To measurethe impact ofpracticereview and in-servicesupervisionbased on WHOguidelines onoutcomes ofseverelymalnourishedchildren in arural facilitywith minimalresourcesstaffed onlyby nurses.

TFC in ruralBailundo,Angola (Jan toAugust 2006)

- Intervention study withhistorical comparison ofoutcomes (January toAugust 2005).

- No control group

Intervention = weeklyphysician supervision ofstaff activities andestablishment of a studygroup composed of nursesin the centre.

- All children aged 0to 12y admitted tothe feeding centreduring the studyperiod with severemalnutrition(defined as WHZ < -3% or <70% ofNCHS referencemedian and/orassociated oedema.

N=379 duringintervention

N= 358 beforeintervention

- Routine practices inthe centre.

- Successfultreatment,withdrawal, death,discharge withoutprogress and transferto another ward werecompared to indicatorsbefore theintervention.

Info before intervention(Jan- Aug 2005)Successfultreatment=73.2%,

Withdrawal=7.8%,

Death= 15.6%,

Discharge withoutprogress= 0.8%,

Transfer to anotherward= 2.5%

- Improved delivery ofimportant tasks such as frequentfeeding and avoidance ofintravenous rehydration wasobserved.

- Among the 379 childrenadmitted during theintervention, compared to the358 children admittedpreviously, successful treatmentincreased from 73.2% to 82.6%(RR 1.13; 95% CI 1.04 to 1.22)and fatalities decreased from15.6% to 8.7% (RR 0.56; 95% CI0.37 to 0.83).Also withdrawals decreased from7.8% to 5.8% (RR 0.74; 95% CI0.43 to 1.27) and dischargewithout progress decreased from0.8% to 0% (RR 0.16; 95% CI 0.01to 3.13).

- This short-termintervention with in-service supervisionbased on the WHOguidelines in asetting of limitedresources apparentlycontributed to areduction in fatalityrates.

These findingssupport the view thatsuch guidelines canbe effectivelyimplemented inunder-resourcedfacilities in Angolaand similar settings ifthey are introducedusing an interactiveapproach and if in-service supervisioncontinues to beprovided.

Gera (2010) To a) assess theefficacy &safety of homebasedmanagementof SAM usingtherapeuticnutritionproducts(TNPs) orRUTFs,b) assess theefficiency ofthese productsin comparisonwith F-100and home-based diet, toextend itsapplication inIndia.

India (April2010)

Systematic review. Electronicdatabase of Pubmed andCochrane Controlled TrialsRegister were scanned,information was extractedfrom identified papers andgraded according to theCEBM guidelines.

Interventions: TNPs, basicallyderived from F-100formulation but should havethese properties;i) does not need to beprepared in any form beforeconsumption,ii) resists microbialcontamination,iii) can be stored at ambienttemperature.

Control: No intervention oralternative intervention suchas home based dietarymanagement or facilitybased standard treatment.

Children withsevere acutemalnutrition.(Age group notdefined)

Trials includedSystematicReviews; beforeand afterobservationalstudies; Individual,cluster, and quasirandomized andnon-randomizedcontrolled trials aswell as consensusstatements.

For efficacy- Recovery rate (asdefined by theauthors) - Weight gain (g/kg/day) - Relapse.

For SafetyMorbidities likediarrhoea, malaria andrespiratory infections

- Mortality.

- Eighteen published papers (2systematic reviews, 7 controlledtrials, 7 observational trials and 2consensus statements) wereidentified.

- Systematic reviews and RCTsshowed RUTF to be at least asefficacious as F-100 in increasingweight (WMD=3.0 g/kg/day; 95%CI -1.70, 7.70) and more effectivein comparison to home baseddietary therapies. Locally madeRUTFs were as effective asimported RUTFs (WMD=0.07g/kg/d; 95% CI=-0.15, 0.29). Datafrom observational studiesshowed the energy intake withRUTF to be comparable to F-100.The pooled recovery rate,mortality and default intreatment with RUTF was 88.3%,0.7% and 3.6%, respectively witha mean weight gain of 3.2g/kg/day. The two consensusstatements supported the use ofRUTF for home basedmanagement of uncomplicatedSAM.

The use of therapeuticnutrition products likeRUTF for home basedmanagement ofuncomplicated SAMappears to be safe andefficacious. However,most of the evidenceon this promisingstrategy has emergedfrom observationalstudies conducted inemergency settings inAfrica. There is need togenerate more robustevidence, designsimilar productslocally and establishtheir efficacy and cost-effectiveness in a‘non-emergency’setting, particularly inthe Indian context.

Mamidi et al(2010)

To examinethe catch upgrowth inseverelywastedchildren usingenergy denselocal foods ata hospitalbasednutritionrehabilitationunit.

Nutritionward in ahospital inHyderabad,India (Jan 2001 toDec 2005)

Retrospective cohort study

Intervention A diet basedon energy dense local foodsalong with multivitamin-multimineral supplements.

- No control group


Children < 5y withsevere wasting(WHZ score < -3) ornutritional oedemaadmitted tonutrition wardN=309

Case sheets ofchildren whostayed for ≥ 7 daysat the nutritionward wereselected.

- Catch up growth(g/kg/day) during eachweek of hospital stay

Baseline infoMean age of thechildren was 25 mo(range 2-60). MeanWHZ was –4.1

Mean weight gain was moderate(5g/kg/day) and baseline WHZscore had a significant negativerelationship to the weight gain.

The prevalence of morbiditieswas high and the commonestmorbidity was fever.

Weight gain was higher byalmost 40% in the absence ofmorbidities in any week.

The diet based onlocal energy densefoods was found to besuitable for thenutritionrehabilitation ofseverelymalnourishedchildren though therate of weight gainwas moderate.

Cont’d next page


61

Authorof study

Objectiveof Study


Indicator /outcome

Impact Conclusion

Akram et al(2010)

To improvenutrition ofmalnourishedchildren in thecommunity,using homebasedtreatment.

Squattersettlement,Karachi,Pakistan (Aug 2006 toMar 2007)

A prospective cohort study

Intervention – locally preparedhigh density diet (HDD) whichprovides 1. 4 calories per ml andhas a shelf life of one week.

- Prepared HDD was taken tochildren’s respective homes bycommunity health workers andfirst feed given under directsupervision.

For the first 4-5 days,100cal/kg/day was provided whichgradually increased to120cal/kg/day over a week.

- No control group

- No statistics presented

All children < 5years who were <-3SD W/H as perWHO classificationN= 24

Children withcomplications werereferred to ahospital for initialstabilization

- Daily weight

- Amount of HDDconsumed

Baseline infoThe mean weightof the childrenwas 7.5 ± 1.84 kg

Eleven children (45.8%)reached -1 SD at the end of 3months while 10 patients(41.6%) took 4 months.Twenty two patients (91.6%)were at the median weight-for-height by the end of 5months.

Home basedtreatment withlocally available foodscan be usedsuccessfully torehabilitate severelymalnourishedchildren.

Patel et al(2010)

To evaluatethe feasibilityand outcomeof home-basedrehabilitationof severelymalnourishedchildren.

Tertiary carehospital inDelhi, NorthIndia.(Nov 2006-Feb 2008)

Prospective and observationalstudy

Intervention -rehabilitation athome (16 weeks) following initialassessment or/and stabilization athospital. Mothers were givenadvice about type of food,quantity of food, and feedingfrequency.

- No external support forprocuring or making food wasprovided to the families. Homebased foods were advised toprovide 150 kcal/kg/day and 2-3g/kg/day of proteins which wasgradually increased so as toprovide approximately 150-220kcal/kg/day and proteins 4-5g/kg/day. Daily multivitaminsupplement was continued till 16weeks.

- Frequency of follow-up visitswere: 2 contacts/ week separatedby at least 48 hours in first twoweeks, then once a week for 3-8weeks and from 8 weeks till 16weeks, every 4 weeks.

- No control group


Severelymalnourished(weight for length<70% of WHOreference) childrenaged 6 mo to 5years.

N= 34 of which 19children wereadmitted inhospital and 15children were senthome after initialassessment inhospital. Five didnot clear the initialstabilization phase(2 died, 3 lefthospital).

- Completerecovery (if childachieved weightfor length >80%at the end of 16weeks).

- Partial recovery(if the weight forlength remainedbetween 70% to80% after 16weeks of follow-up).

- Weight gain

Baseline infoIncorporated inresults.Out of 29 childrenwho qualified forhome basedrehabilitation, 26completed 16week follow-up.

During the home basedmanagement phase, thereported mean (±SD) calorieintake increased from 100 (±5)kcal/kg/d at entry point to 243(±13) kcal/kg/d at 16 weeks(P=0.000).Similarly, reported proteinintake increased from 1.1(±0.3) g/kg/d to 4.8 (±0.3)g/kg/d (P=0.000). Duringhospital stay (n=19), childrenhad weight gain of 9.0 (±5.3)g/kg/d, while during homebased follow up (n=29),weight gain was 3.2 (±1.5)g/kg/d only. During homebased rehabilitation, only3 (11.5%) children had weightgain of more than 5 g/kg/d bythe end of 16 weeks. W/H%increased from an average of62.9% (±6.0%) to 80.3%(±5.7%) after the completionof 16 weeks (P=0.000).Thirteen (45%) childrenrecovered completely frommalnutrition whereas 15(51.7%) recovered partially.

There was no death during thehome stabilization.

Home basedmanagement usinghome prepared foodand hospital basedfollow up isassociated with sub-optimal and slowerrecovery.

Trehan et al(2010)

To determinewhether theinclusion ofamoxicillincorrelateswith betterrecovery ratesin the home-basedtreatment ofsevere acutemalnutritionwith ready-to-usetherapeuticfood.

Malnutritionclinics atDowa,Chiradzulu,and Machingadistricts,Malawi (2003-2005)

- Retrospective cohort studycomparing two groups of childrenwith uncomplicated SAM.

- The standard protocol groupreceived a 7-day course ofamoxicillin at the onset oftreatment while the alternateprotocol group received noantibiotics. All children weretreated with the same ready-to-use therapeutic food up to a totalof 12 weeks.Reassessment was done every 1 to2 weeks.

- Control group


Children withuncomplicated SAM(WHZ ≤ -3 of WHOreference and ⁄orthe presence ofbilateral pittingoedema) aged 6-59months.

N= 2543 of whichstandard protocolgroup= 498 andalternate protocolgroup= 1955

- Recovery rate(defined asachieving WHZ > -2 withoutoedema).

Baseline infoMean ± SD WHZwas 1.99 ± 1.26and 1.91 ± 1.45 inthe standardprotocol groupand alternateprotocol grouprespectively.

Children in thestandard protocolgroup wereslightly older andmore stunted.

The recovery rate for childrenwho received amoxicillin wasworse at 4 weeks (40% vs.71%) but similar after up to 12weeks of therapy (84% vs.86%), compared to thechildren treated withoutantibiotics. Regressionmodelling indicated that thisdifference at 4 weeks was moststrongly associated with thereceipt of amoxicillin.

This review of twotherapeutic feedingprogrammes suggeststhat children withsevere acutemalnutrition whowere treated withoutamoxicillin did nothave an inferior rateof recovery. Given thelimitations of thisretrospective analysis,a prospective trial iswarranted todetermine the effectof antibiotics onrecovery fromuncomplicatedmalnutrition withhome-based therapy.

Cont’d next page


62

Authorof study

Objectiveof Study


Indicator /outcome

Impact Conclusion

Oakley et al(2010)

To determinewhethertreatingchildren withSAM with 10%milk RUTFcontainingsoy wouldresult in asimilarrecovery ratecomparedwith the 25%milk RUTF.

Fifteen ruralsites insouthernregion ofMalawi(Jul 2008 toApr 2009)

A randomised, double-blind,clinical, quasi-effectiveness trialof 2 locally produced foods forthe treatment of SAM.

Children were randomlyassigned with equal probabilityto either 25% milk RUTF or 10%milk RUTF as home-basedtherapy for up to 8 weeks.

Children were assessed everyfortnight and participated inthe study until they clinicallyrecovered or received 8 wk oftreatment.

- No control group


Severelymalnourished(WHZ < -3 and/orhaving bipedalpitting oedema,WHO reference)children aged 6-59mo with a goodappetite

N= 1874 of which25% milkRUTFgroup= 945and 10% milkgroup= 929

Recovery (defined asWHZ > -2 and nooedema)

Weight and heightgain

Baseline infoMean (SD) WHZ was -2.1(1.2) and -2.0(1.2)in the 25% milk and10% milk RUTF groupsrespectively.

Recovery among childrenreceiving 25% milk RUTF wasgreater than childrenreceiving 10% milk RUTF,64% compared with 57%after 4 wk, and 84%compared with 81% after 8wk (p< 0.001)

Children receiving 25% milkRUTF also had higher rates ofweight and height gaincompared with childrenreceiving 10% milk RUTF.

Treating childrenwith SAM with 10%milk RUTF is lesseffective comparedwith treatment withthe standard 25%milk RUTF.These findings alsoemphasize thatclinical evidenceshould be examinedbeforerecommending anychanges to theformulation of RUTF.


63

Authorof study

Objective ofStudy


Surveypopulation

Indicator /outcome

Impact Conclusion

Smuts et al(2005)

To investigatewhetherimproving statusfor multiplemicronutrientsprevented growthfaltering andanaemia duringinfancy.

(The Four CountryIRIS Trial PooledData Analysis)

Indonesia,Peru, SouthAfrica, andVietnam(dates givenelsewhere)

A randomized placebocontrolled trial.

Participants were randomizedto 4 treatment groups; dailyplacebo (P), weekly multiplemicronutrient supplement(WMM), daily multiplemicronutrient (DMM)supplement, and daily iron (DI)supplements.

Each child had 2 measurements,at baseline (pre) and 6 mo(post) intervention.

- Control group


Infants aged 6-11monot bornprematurely (<37wk gestation), notborn low birthweight (<2500 g),not severely wasted(< -3 Z-scores), notseverely anaemic(haemoglobin <80g/L), and no fever (>39°C).N= 1134 of which P group= 283, WMMgroup= 283, DMMgroup= 280 and DIgroup= 288.

Weight gain

Height gain

Anaemiaprevalence

Baseline infoThe prevalence ofmultiplemicronutrientdeficiencies atbaseline was high,with anaemiaaffecting themajority,

The DMM group had asignificantly greater weight gain(p= 0.036), growing at anaverage rate of 207 g/mocompared with 192 g/mo for theWMM group, and 186 g/mo forthe DI and P groups.

There were no differences inheight gain.

DMM was also the most effectivetreatment for controllinganaemia and iron deficiency,besides improving zinc, retinol,tocopherol, and riboflavin status.

DI supplementation aloneincreased zinc deficiency.

The prevalence of anaemia wasnot fully controlled even after 6mo of supplementation.

These positiveresults indicate theneed for largereffectiveness trialsto examine how todeliver supplementsat the programmescale and toestimate costbenefits.Considerationshould also be givento increasing thedosages ofmicronutrientsbeing delivered inthe foodlets.

Giovanniniet al (2006)

To assess andcompare efficacyof twomicronutrientsprinklesupplementationon growth,anaemia, andiron deficiency inCambodianinfants.

KompongChhnangProvince,Cambodia

- Double-blind, placebo-controlled, comparativelongitudinal trial

- Participants were randomlyassigned to receive dailysupplements of either iron(12.5 mg) plus folic acid (150Kg) plus zinc (5 mg) (MMN), oriron (12.5 mg) plus folic acid(150 Kg) alone (FFA), orplacebo for a 12 month periodin powder form as sprinkles

- Reassessment wasperformed weekly

- Control group


Infants aged 6months withhaemoglobin ≥ 70g/L.N= 204 of whichMMN group = 68,FFA group= 68placebo group= 68

Growth pattern

Anaemia

Iron deficiency

Baseline infoMean (SD; range)of Hb (g/L) was99.4 (9.9; 76-119)101.0 (10.1; 81-128) and 102.9(10.2; 80-123) forMMN, FFA andplacebo groupsrespectively.

No difference among groupswas found for growth pattern.Significant decline was observedfor WAZ and HAZ in any group (P< 0.0001).

The rate of recovery fromanaemia was significant (P <0.001) and comparable betweenMMN (54%) and FFA (53%)groups and higher than in theplacebo group (22%, P <0.0001).

Through the study period, nosignificant change in the rate ofiron deficiency was found inMMN and FFA groups, whereas itincreased in the placebo group(31%, baseline vs. 52%, end ofstudy; P < 0.0001).

Both MMN and FFAsupplements wereeffective forpreventing ortreating anaemia inCambodian infantsand stabilizingplasma levels offerritin. Use ofmicronutrients in acontrolled homesetting, as sprinkleddaily supplements,may be promising inpreventing andtreating anemia indevelopingcountries.

Ip et al(2009)

To compare theeffectiveness ofdaily and flexibleadministration ofmicronutrientSprinkles onadherence,acceptability andhaematologicalstatus amongyoung children inruralBangladesh.

the Kaliganjsub-district ofGazipurdistrictBangladesh(May - Sep2004)

A cluster randomised trial

Children were cluster-randomized according to theirvillage (cluster) to receive 60sachets of Sprinkles either (i) daily over 2 months; (ii) flexibly over 3 months; or (iii) flexibly over 4 months. With a flexible regimen,mothers/caregivers decidedhow frequently to useSprinkles without exceedingone sachet per day. Adherencewas assessed monthly bycounting the number ofsachets used and acceptabilitywas evaluated through focusgroup discussions.Haemoglobin was measured atbaseline, at the end of eachintervention period and 6months post-intervention.

- No control group


Children aged 6–24months withhaemoglobin ≥70g/l.N= 362 childrenfrom 16 villages

Group adherence

Hb levels

Anaemia (definedas Hb <110 g /l)prevalence

Baseline infoGender, mean Hband anaemiaprevalence weresimilar among thethree groups atbaseline.Overall, mean Hbwas 99 ± 14 g /land 278/362(77%) childrenwere anaemic andprevalence ofstunting (29%),underweight(41.7%) andwasting (13.8%).

Mean percent adherence wassignificantly higher in theflexible-4-month group (98%)compared to the flexible- 3-month (93%) and daily-2-month(88%) groups (P< 0.01). Most mothers found flexibleadministration to be moreacceptable than daily due toperceived benefits of use. Hb atthe end of intervention wassignificantly higher in theflexible-4-month groupcompared to the daily group(P= 0.03). Anaemia prevalence decreasedby 65% in the flexible-4-monthgroup compared to 54% in theflexible-3-month and 51% inthe daily-2-month groups.Percent of cured children whomaintained a non-anaemicstatus 6 months post-intervention was significantlyhigher in the flexible-4-month(82%) and flexible-3-month(80%) groups than the daily-2-month (53%) group (P<0.05).

The adherence,acceptability andhaematologicalresponse to flexibleadministration over4 months werefound preferable todaily.

5. Summary of published studies assessing the impact of micronutrients distribution programmes in non-emergencies

The impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidence References

64

References10

1. WHO (1995). Physical status: the use and interpretation of anthropometry. Geneva: World Health Organization.2. Watson, F., J. Shoham, C. Dolan and A. Hall (2011). The quality and potential use of data collected during nutrition surveys:

an analysis of surveys in Ethiopia. International Health, 3 (2): 85-90.3. Lucas, R.M. and A.J. McMichael (2005). Association or causation: evaluating links between "environment and disease". Bull

World Health Organ, 83 (10): 792-795.4. Duffield, A., G. Reid, D. Walker and J. Shoham, Review of the published literature for the impact and cost-effectiveness of six

nutrition related emergency interventions. 2004, Emergency Nutrition Network: Oxford. http://www.ennonline.net/pool/files/ife/enncidasfpreport.pdf

5. Porta, M. (2008). A Dictionary of Epidemiology. 5th. ed. Oxford: Oxford University Press.6. Hipgrave, D.B., F. Assefa, A. Winoto and S. Sukotjo (2011). Donated breast milk substitutes and incidence of diarrhoea

among infants and young children after the May 2006 earthquake in Yogyakarta and Central Java. Public Health Nutr: 1-9.7. Habicht, J.P., C.G. Victora and J.P. Vaughan (1999). Evaluation designs for adequacy, plausibility and probability of public

health programme performance and impact. Int J Epidemiol, 28 (1): 10-18.8. Bierce, A. (1911). The Devil's Dictionary: http://www.alcyone.com/max/lit/devils/.9. Chotard, S., J.B. Mason, N.P. Oliphant, S. Mebrahtu and P. Hailey (2010). Fluctuations in wasting in vulnerable child

populations in the Greater Horn of Africa. Food Nutr Bull, 31 (3 Suppl): S219-233.10. Lagarde, M. (2011). How to do (or not to do) ... Assessing the impact of a policy change with routine longitudinal data.

Health Policy Plan.11. Hall, A., M. Oirere, S. Thurstans, A. Ndumi and V. Sibson (2011). An attempt to evaluate a blanket emergency feeding

programme in Kenya. Submitted.12. Mamidi, R.S., B. Kulkarni, K.V. Radhakrishna and V. Shatrugna (2010). Hospital based nutrition rehabilitation of severely

undernourished children using energy dense local foods. Indian Pediatr, 47 (8): 687-693.13. Collins, S., K. Sadler, N. Dent, T. Khara, S. Guerreroet al. (2006). Key issues in the success of community-based management

of severe malnutrition. Food Nutr Bull, 27 (3 Suppl): S49-82.14. Matilsky, D.K., K. Maleta, T. Castleman and M.J. Manary (2009). Supplementary feeding with fortified spreads results in

higher recovery rates than with a corn/soy blend in moderately wasted children. J Nutr, 139 (4): 773-778.

ACF-USA, Kenya, 2009

ReferencesThe impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidence

65

15. Nackers, F., F. Broillet, D. Oumarou, A. Djibo, V. Gaboulaudet al. (2010). Effectiveness of ready-to-use therapeutic food compared to a corn/soy-blend-based pre-mix for the treatment of childhood moderate acute malnutrition in Niger. J Trop Pediatr, 56 (6): 407-413.

16. Ndekha, M.J., M.J. Manary, P. Ashorn and A. Briend (2005). Home-based therapy with ready-to-use therapeutic food is of benefit to malnourished, HIV-infected Malawian children. Acta Paediatr., 94: 222-225.

17. Collins, S. and K. Sadler (2002). Outpatient care for severely malnourished children in emergency relief programmes: a retrospective cohort study. Lancet, 360 (9348): 1824-1830.

18. The Sphere Project, The Sphere Project. Humanitarian charter and minimum standards in humanitarian response. 2011, Practical Action Publishing: Rugby, UK. p. 393.

19. Jayatissa, R., A. Bekele, C.L. Piyasena and S. Mahamithawa (2006). Assessment of nutritional status of children under five years of age, pregnant women, and lactating women living in relief camps after the tsunami in Sri Lanka. Food Nutr Bull, 27 (2): 144-152.

20. Renzaho, A.M. (2007). Mortality rates, prevalence of malnutrition, and prevalence of lost pregnancies among the drought-ravaged population of Tete Province, Mozambique. Prehosp Disaster Med, 22 (1): 26-34.

21. Johnecheck, W.A. and D.E. Holland (2007). Nutritional status in postconflict Afghanistan: evidence from the National Surveillance System Pilot and National Risk and Vulnerability Assessment. Food Nutr Bull, 28 (1): 3-17.

22. Olwedo, M.A., E. Mworozi, H. Bachou and C.G. Orach (2008). Factors associated with malnutrition among children in internally displaced person's camps, northern Uganda. Afr Health Sci, 8 (4): 244-252.

23. Rossi, L., T. Hoerz, V. Thouvenot, G. Pastore and M. Michael (2006). Evaluation of health, nutrition and food security programmes in a complex emergency: the case of Congo as an example of a chronic post-conflict situation. Public Health Nutr, 9 (5): 551-556.

24. Shrimpton, R., A. Thorne-Lyman, K. Tripp and A. Tomkins (2009). Trends in low birthweight among the Bhutanese refugee population in Nepal. Food Nutr Bull, 30 (2 Suppl): S197-206.

25. Kumar, S. and L. Bhawani (2005). Managing child malnutrition in a drought affected district of Rajasthan--a case study. Indian J Public Health, 49 (4): 198-206.

26. Svedberg, P. (2000). Poverty and undernutrition. Theory, measurement and policy. Oxford: Oxford University Press.27. Amthor, R.E., S.M. Cole and M.J. Manary (2009). The use of home-based therapy with ready-to-use therapeutic food to treat

malnutrition in a rural area during a food crisis. J Am Diet Assoc, 109 (3): 464-467.28. Ciliberto, M.A., H. Sandige, M.J. Ndekha, P. Ashorn, A. Briendet al. (2005). Comparison of home-based therapy with ready-to-

use therapeutic food with standard therapy in the treatment of malnourished Malawian children: a controlled, clinical effectiveness trial. Am J Clin Nutr, 81 (4): 864-870.

29. Ciliberto, M.A., M.J. Manary, M.J. Ndekha, A. Briend and P. Ashorn (2006). Home-based therapy for oedematous malnutritionwith ready-to-use therapeutic food. Acta Paediatr, 95 (8): 1012-1015.

30. Linneman, Z., D. Matilsky, M. Ndekha, M.J. Manary and K. Maleta (2007). A large-scale operational study of home-based therapy with ready-to-use therapeutic food in childhood malnutrition in Malawi. Matern Child Nutr, 3 (3): 206-215.

31. Manary, M.J., M.J. Ndkeha, P. Ashorn, K. Maleta and A. Briend (2004). Home based therapy for severe malnutrition with ready-to-use food. Arch Dis Child, 89 (6): 557-561.

32. Sandige, H., M.J. Ndekha, A. Briend, P. Ashorn and M.J. Manary (2004). Home-based treatment of malnourished Malawian children with locally produced or imported ready-to-use food. J Pediatr Gastroenterol Nutr, 39 (2): 141-146.

33. Fjeld, C.R. and D.A. Schoeller (1988). Energy expenditure of malnourished children during catch-up growth. Proc Nutr Soc, 47 (3): 227-231.

34. Khanum, S., A. Ashworth and S.R. Huttly (1994). Controlled trial of three approaches to the treatment of severe malnutrition. Lancet, 344 (8939-8940): 1728-1732.

35. Ashraf, H., T. Ahmed, M.I. Hossain, N.H. Alam, R. Mahmudet al. (2007). Day-care management of children with severe malnutrition in an urban health clinic in Dhaka, Bangladesh. J Trop Pediatr, 53 (3): 171-178.

36. Patel, M.P., H.L. Sandige, M.J. Ndekha, A. Briend, P. Ashornet al. (2005). Supplemental feeding with ready-to-use therapeutic food in Malawian children at risk of malnutrition. J Health Popul Nutr, 23 (4): 351-357.

37. Maleta, K., J. Kuittinen, M.B. Duggan, A. Briend, M. Manaryet al. (2004). Supplementary feeding of underweight, stunted Malawian children with a ready-to-use food. J Pediatr Gastroenterol Nutr, 38 (2): 152-158.

38. Thakwalakwa, C., P. Ashorn, J. Phuka, Y.B. Cheung, A. Briendet al. (2010). A lipid-based nutrient supplement but not corn-soyblend modestly increases weight gain among 6- to 18-month-old moderately underweight children in rural Malawi. J Nutr, 140 (11): 2008-2013.

39. Nielsen, J., P. Valentiner-Branth, C. Martins, F. Cabral and P. Aaby (2004). Malnourished children and supplementary feeding during the war emergency in Guinea-Bissau in 1998-1999. Am J Clin Nutr, 80 (4): 1036-1042.

The impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidence References

66

40. Ruel, M.T., P. Menon, J.P. Habicht, C. Loechl, G. Bergeronet al. (2008). Age-based preventive targeting of food assistance and behaviour change and communication for reduction of childhood undernutrition in Haiti: a cluster randomised trial. Lancet, 371 (9612): 588-595.

41. Defourny, I., A. Minetti, G. Harczi, S. Doyon, S. Shepherdet al. (2009). A large-scale distribution of milk-based fortified spreads: evidence for a new approach in regions with high burden of acute malnutrition. PLoS One, 4 (5): e5455.

42. Isanaka, S., N. Nombela, A. Djibo, M. Poupard, D. Van Beckhovenet al. (2009). Effect of preventive supplementation with ready-to-use therapeutic food on the nutritional status, mortality, and morbidity of children aged 6 to 60 months in Niger: a cluster randomized trial. JAMA, 301 (3): 277-285.

43. Greco, L., J. Balungi, K. Amono, R. Iriso and B. Corrado (2006). Effect of a low-cost food on the recovery and death rate of malnourished children. J Pediatr Gastroenterol Nutr, 43 (4): 512-517.

44. Gaboulaud, V., N. Dan-Bouzoua, C. Brasher, G. Fedida, B. Gergonneet al. (2007). Could nutritional rehabilitation at home complement or replace centre-based therapeutic feeding programmes for severe malnutrition? J Trop Pediatr, 53 (1): 49-51.

45. Dubray, C., S.A. Ibrahim, M. Abdelmutalib, P.J. Guerin, F. Dantoineet al. (2008). Treatment of severe malnutrition with 2-day intramuscular ceftriaxone vs 5-day amoxicillin. Ann Trop Paediatr, 28 (1): 13-22.

46. Sadler, K., M. Myatt, T. Feleke and S. Collins (2007). A comparison of the programme coverage of two therapeutic feeding interventions implemented in neighbouring districts of Malawi. Public Health Nutr, 10 (9): 907-913.

47. Rossi, L., D. Verna and S.L. Villeneuve (2008). The humanitarian emergency in Burundi: evaluation of the operational strategy for management of nutritional crisis. Public Health Nutr, 11 (7): 699-705.

48. Guerrero, S., M. Myatt and S. Collins (2010). Determinants of coverage in community-based therapeutic care programmes: towards a joint quantitative and qualitative analysis. Disasters, 34 (2): 571-585.

49. Seal, A.J., P.I. Creeke, Z. Mirghani, F. Abdalla, R.P. McBurneyet al. (2005). Iron and vitamin A deficiency in long-term African refugees. J Nutr, 135 (4): 808-813.

50. Seal, A.J., P.I. Creeke, D. Gnat, F. Abdalla and Z. Mirghani (2006). Excess dietary iodine intake in long-term African refugees. Public Health Nutr, 9 (1): 35-39.

51. Lopriore, C., Y. Guidoum, A. Briend and F. Branca (2004). Spread fortified with vitamins and minerals induces catch-up growth and eradicates severe anemia in stunted refugee children aged 3-6 y. Am. J. Clin. Nutr., 80: 973-981.

52. Moench-Pfanner, R., S. de Pee, M.W. Bloem, D. Foote, S. Kosenet al. (2005). Food-for-work programmes in Indonesia had a limited effect on anemia. J Nutr, 135 (6): 1423-1429.

53. Nielsen, J., C.S. Benn, C. Bale, C. Martins and P. Aaby (2005). Vitamin A supplementation during war-emergency in Guinea-Bissau 1998-1999. Acta Trop, 93 (3): 275-282.

54. Menon, P., M.T. Ruel, C.U. Loechl, M. Arimond, J.P. Habichtet al. (2007). Micronutrient Sprinkles reduce anemia among 9- to 24-mo-old children when delivered through an integrated health and nutrition programme in rural Haiti. J Nutr, 137 (4): 1023-1030.

55. World Health Organization, Iron deficiency anaemia. Assessment, prevention and control. A guide for programme managers. 2001, World Health Organization: Geneva.

56. Seal, A., E. Kafwembe, I.A. Kassim, M. Hong, A. Wesleyet al. (2008). Maize meal fortification is associated with improved vitamin A and iron status in adolescents and reduced childhood anaemia in a food aid-dependent refugee population. Public Health Nutr, 11 (7): 720-728.

57. Magoni, M., M. Jaber and R. Piera (2008). Fighting anaemia and malnutrition in Hebron (Palestine): impact evaluation of a humanitarian project. Acta Trop, 105 (3): 242-248.

58. Talley, L., B.A. Woodruff, A. Seal, K. Tripp, L.S. Mselleet al. (2010). Evaluation of the effectiveness of stainless steel cooking pots in reducing iron-deficiency anaemia in food aid-dependent populations. Public Health Nutr, 13 (1): 107-115.

59. Adish, A.A., S.A. Esrey, T.W. Gyorkos, J. Jean-Baptiste and A. Rojhani (1999). Effect of consumption of food cooked in iron pots on iron status and growth of young children: a randomised trial. Lancet, 353 (9154): 712-716.

60. Borigato, E.V. and F.E. Martinez (1998). Iron nutritional status is improved in Brazilian preterm infants fed food cooked in iron pots. J Nutr, 128 (5): 855-859.

61. WHO/WFP/UNICEF, Preventing and controlling micronutrient deficiencies in populations affected by an emergency. n.d., World Health Organization: Geneva.

62. Bushamuka, V.N., S. de Pee, A. Talukder, L. Kiess, D. Panagideset al. (2005). Impact of a homestead gardening programme onhousehold food security and empowerment of women in Bangladesh. Food Nutr Bull, 26 (1): 17-25.

63. De Waal, A., A.S. Taffesse and L. Carruth (2006). Child survival during the 2002-2003 drought in Ethiopia. Glob Public Health, 1 (2): 125-132.

64. Mascie-Taylor, C., M. Marks, R. Goto and R. Islam (2010). Impact of a cash-for-work programme on food consumption and nutrition among women and children facing food insecurity in rural Bangladesh. Bull World Health Organ, 88 (11): 854-860.

65. Morris, S.S., P. Olinto, R. Flores, E.A. Nilson and A.C. Figueiro (2004). Conditional cash transfers are associated with a small reduction in the rate of weight gain of preschool children in northeast Brazil. J Nutr, 134 (9): 2336-2341.

ReferencesThe impact and effectiveness of emergency nutrition and nutrition-related interventions: a review of published evidence

67

66. Barber, S.L. and P.J. Gertler (2008). The impact of Mexico's conditional cash transfer programme, Oportunidades, on birthweight. Trop Med Int Health, 13 (11): 1405-1414.

67. Lagarde, M., A. Haines and N. Palmer (2009). The impact of conditional cash transfers on health outcomes and use of health services in low and middle income countries. Cochrane Database Syst Rev, (4): CD008137.

68. Abebe, D., A. Cullis, A. Catley, Y. Aklilu, G. Mekonnenet al. (2008). Impact of a commercial destocking relief intervention in Moyale district, southern Ethiopia. Disasters, 32 (2): 167-189.

69. Gelli, A., N. Al-Shaiba and F. Espejo (2009). The costs and cost-efficiency of providing food through schools in areas of high food insecurity. Food Nutr Bull, 30 (1): 68-76.

70. Hall, A., G. Hewitt, V. Tuffrey and N. de Silva (2008). A review and meta-analysis of the impact of intestinal worms on child growth and nutrition. Matern Child Nutr, 4 Suppl 1: 118-236.

71. Emergency Nutrition Network. (Year). WHO, UNICEF, WFP and UNHCR consultation on the programmatic aspects of the management of moderate acute malnutrition in children under five years of age. 24-26 February 2010, WHO, Geneva. Geneva: Emergency Nutrition Network.

72. Inter-agency Food and Nutrition Group. (Year). Report of a Workshop on Selective Feeding Programmes. Dublin, Ireland, 18th-20th 1997. 1997. Dublin: Concern Worldwide.

73. Ockwell, R. (Year). Food security and food assistance among long-standing refugees. 1999: WFP/UNHCR 74. Emergency Nutrition Network. (Year). Community based approaches to managing severe malnutrition. Report on the

proceedings of an inter-agency workshop. Dublin 8th-10th October, 2003. 2003: Emergency Nutrition Network.75. FANTA/ENN. (Year). International workshop on the integration of community-based management of acute malnutrition,

Washington DC, April 28 - 30, 2008. Workshop Report. 2008. Washington, DC: FANTA/ENN.76. Reed, B.A., J.P. Habicht and C. Garza (2002). Translating nutrition research into action in humanitarian emergencies. J Nutr,

132 (7): 2112S-2116S.77. Walker, S.P., S.M. Chang, C.A. Powell and S.M. Grantham-McGregor (2005). Effects of early childhood psychosocial

stimulation and nutritional supplementation on cognition and education in growth-stunted Jamaican children: prospective cohort study. Lancet, 366 (9499): 1804-1807.

78. Hamadani, J.D., S.N. Huda, F. Khatun and S.M. Grantham-McGregor (2006). Psychosocial stimulation improves the development of undernourished children in rural Bangladesh. J Nutr, 136 (10): 2645-2652.

79. Walker, S.P., S.M. Chang, C.A. Powell, E. Simonoff and S.M. Grantham-McGregor (2006). Effects of psychosocial stimulation and dietary supplementation in early childhood on psychosocial functioning in late adolescence: follow-up of randomised controlled trial. BMJ, 333 (7566): 472.

Emergency Nutrition Network (ENN)32, Leopold Street, Oxford, OX4 1TW, UK

Tel: +44 (0)1865 324996Fax: +44 (0)1865 324997Email: [email protected]

the impact and effectiveness of emergency nutrition and ... · therapeutic nutrition programmes and...

Documents